{
  "benchmark_version": "0.3.0",
  "experiment_id": "weekly-board",
  "pricing_version": "2026-07-09",
  "cost_basis": "billed",
  "baseline_model_key": "or:gpt-oss-120b",
  "calculated_at": "2026-08-22T15:48:07.400963",
  "task_basket": [
    "banking-day-count-convention@1",
    "banking-effective-annual-rate@1",
    "banking-fx-missing-rate-refusal@1",
    "banking-kyc-risk-tier@1",
    "banking-payment-waterfall@1",
    "banking-statement-field-extraction@1",
    "coding-add-months@1",
    "coding-dedup-implementation@1",
    "coding-dedup-preserve-order@1",
    "coding-merge-intervals@1",
    "coding-pair-sums-performance@1",
    "coding-parse-duration-contract@1",
    "coding-subtract-intervals@1",
    "cs-defective-return-exception@1",
    "cs-escalation-decision@1",
    "cs-prorated-refund@1",
    "cs-refund-eligibility@1",
    "cs-ticket-intent-classification@1",
    "legal-clause-precedence@1",
    "legal-governing-law-extraction@1",
    "legal-liability-cap-carveout@1",
    "legal-notice-period-calculation@1",
    "legal-termination-earliest-date@1",
    "medical-conditional-interaction@1",
    "medical-icd10-lookup@2",
    "medical-infusion-rate-units@1",
    "medical-interaction-extraction@1",
    "medical-max-dose-cap@1",
    "medical-missing-weight-refusal@1",
    "medical-paediatric-dose-volume@1",
    "medical-reference-range-interpretation@1",
    "medical-renal-dose-adjustment@1"
  ],
  "snapshot_count": 1292,
  "snapshots": [
    {
      "snapshot_id": "ae8db714-9cb3-4095-9cb4-a91367d8e68e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 75,
      "content_successes": 81,
      "success_rate": 0.78125,
      "total_cost_usd": 0.047443,
      "cost_per_success_usd": 0.0008246831475287998,
      "octane_score": 8.53638857645015,
      "ci_low": 2.758832286758213,
      "ci_high": 17.90176739719174,
      "ci_metric": "octane_score",
      "median_latency_ms": 716.7656249366701,
      "p95_latency_ms": 3168.655312154442,
      "input_tokens": 11808,
      "output_tokens": 7127,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f9326a0a-2560-4ac3-aba4-2741c3429cdb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 92,
      "content_successes": 95,
      "success_rate": 0.9583333333333334,
      "total_cost_usd": 0.302115,
      "cost_per_success_usd": 0.0033568375345065563,
      "octane_score": 2.0971571389410806,
      "ci_low": 1.2059582685453645,
      "ci_high": 3.597886681754457,
      "ci_metric": "octane_score",
      "median_latency_ms": 2334.882916882634,
      "p95_latency_ms": 5979.975896188989,
      "input_tokens": 11808,
      "output_tokens": 9723,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e64ee29e-d3a9-4801-adff-aa1a2e9c2a60",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 91,
      "content_successes": 91,
      "success_rate": 0.9479166666666666,
      "total_cost_usd": 0.157659,
      "cost_per_success_usd": 0.0018794827329192548,
      "octane_score": 3.745613448026415,
      "ci_low": 2.336483802335322,
      "ci_high": 5.704958965119892,
      "ci_metric": "octane_score",
      "median_latency_ms": 1878.1898957677186,
      "p95_latency_ms": 5261.205155402422,
      "input_tokens": 11808,
      "output_tokens": 8149,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c0f39475-3d56-4fbf-a5b0-c067d7790ccc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 83,
      "content_successes": 83,
      "success_rate": 0.8645833333333334,
      "total_cost_usd": 0.00496548,
      "cost_per_success_usd": 7.322402019034778e-05,
      "octane_score": 96.14079889980928,
      "ci_low": 46.303032761812915,
      "ci_high": 191.5454984793993,
      "ci_metric": "octane_score",
      "median_latency_ms": 1335.1498330011964,
      "p95_latency_ms": 7914.711790857837,
      "input_tokens": 10836,
      "output_tokens": 5850,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9df02724-3450-44f9-80a8-48dd6d7d0c61",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 85,
      "content_successes": 85,
      "success_rate": 0.8854166666666666,
      "total_cost_usd": 0.05783952,
      "cost_per_success_usd": 0.0008510907306101308,
      "octane_score": 8.271522114580064,
      "ci_low": 1.6278572729651082,
      "ci_high": 137.9811618974109,
      "ci_metric": "octane_score",
      "median_latency_ms": 2645.7454999908805,
      "p95_latency_ms": 100213.2486354094,
      "input_tokens": 13785,
      "output_tokens": 16009,
      "reasoning_tokens": 299339,
      "undefined_reason": null
    },
    {
      "snapshot_id": "218b4d21-f565-4e83-b75f-792e29db0a1d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 85,
      "content_successes": 85,
      "success_rate": 0.8854166666666666,
      "total_cost_usd": 0.352672886,
      "cost_per_success_usd": 0.004922844402005649,
      "octane_score": 1.430030125853188,
      "ci_low": 0.3514268223320246,
      "ci_high": 22.98096001228466,
      "ci_metric": "octane_score",
      "median_latency_ms": 3899.7935210354626,
      "p95_latency_ms": 188397.50509290025,
      "input_tokens": 15826,
      "output_tokens": 7247,
      "reasoning_tokens": 390465,
      "undefined_reason": null
    },
    {
      "snapshot_id": "083cd4f2-281f-409f-a8ce-af74e970ff28",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 84,
      "content_successes": 84,
      "success_rate": 0.875,
      "total_cost_usd": 0.0261465,
      "cost_per_success_usd": 0.0003215450613776467,
      "octane_score": 21.893714584182998,
      "ci_low": 10.860607286577425,
      "ci_high": 74.43717395051694,
      "ci_metric": "octane_score",
      "median_latency_ms": 703.4844374284148,
      "p95_latency_ms": 3458.669406827539,
      "input_tokens": 10605,
      "output_tokens": 9186,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e056f6d2-99b3-41ed-9ab0-7c9e8995973b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 89,
      "content_successes": 89,
      "success_rate": 0.9270833333333334,
      "total_cost_usd": 0.5366475,
      "cost_per_success_usd": 0.006539020796980798,
      "octane_score": 1.076585626246399,
      "ci_low": 0.6676867210230203,
      "ci_high": 1.8045410923135814,
      "ci_metric": "octane_score",
      "median_latency_ms": 2996.3783537968993,
      "p95_latency_ms": 10841.410583117977,
      "input_tokens": 10660,
      "output_tokens": 4073,
      "reasoning_tokens": 65348,
      "undefined_reason": null
    },
    {
      "snapshot_id": "860040ee-c8a3-4e09-b5c4-81dd3541f67d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 92,
      "content_successes": 92,
      "success_rate": 0.9583333333333334,
      "total_cost_usd": 0.085275,
      "cost_per_success_usd": 0.0009875237872344394,
      "octane_score": 7.128755672276823,
      "ci_low": 4.843753758079303,
      "ci_high": 11.435866476019816,
      "ci_metric": "octane_score",
      "median_latency_ms": 3103.1069583259523,
      "p95_latency_ms": 8161.62445791997,
      "input_tokens": 10605,
      "output_tokens": 4126,
      "reasoning_tokens": 39233,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3df7ae73-33bf-497d-ab30-26aac7694fc9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 74,
      "content_successes": 74,
      "success_rate": 0.7708333333333334,
      "total_cost_usd": 0.08603692,
      "cost_per_success_usd": 0.0016197279544887077,
      "octane_score": 4.346295178919738,
      "ci_low": 1.3618253139837813,
      "ci_high": 18.73783866668591,
      "ci_metric": "octane_score",
      "median_latency_ms": 11450.20543737337,
      "p95_latency_ms": 228332.64315640554,
      "input_tokens": 10127,
      "output_tokens": 17079,
      "reasoning_tokens": 196760,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d6988a0e-d22a-4e43-86d0-a190b4a8a0aa",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 79,
      "content_successes": 79,
      "success_rate": 0.8229166666666666,
      "total_cost_usd": 0.46100712,
      "cost_per_success_usd": 0.009543327193650792,
      "octane_score": 0.7376689132527504,
      "ci_low": 0.18000333491350107,
      "ci_high": 4.4016139470915805,
      "ci_metric": "octane_score",
      "median_latency_ms": 11315.116812475026,
      "p95_latency_ms": 269367.8439585492,
      "input_tokens": 10446,
      "output_tokens": 6221,
      "reasoning_tokens": 231003,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7bd5eef2-6707-4a18-bd60-8c75b3a941e2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 91,
      "content_successes": 91,
      "success_rate": 0.9479166666666666,
      "total_cost_usd": 0.2519228364,
      "cost_per_success_usd": 0.0031222857351320887,
      "octane_score": 2.2546994083672423,
      "ci_low": 0.8465876514165992,
      "ci_high": 7.4998387125633075,
      "ci_metric": "octane_score",
      "median_latency_ms": 3011.4426044747233,
      "p95_latency_ms": 28037.031572544947,
      "input_tokens": 10929,
      "output_tokens": 4268,
      "reasoning_tokens": 77158,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1f81cb8e-996d-4422-9889-af234156675d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 85,
      "content_successes": 93,
      "success_rate": 0.8854166666666666,
      "total_cost_usd": 0.17464124,
      "cost_per_success_usd": 0.002268401139571994,
      "octane_score": 3.1034263195107044,
      "ci_low": 1.5428639570062672,
      "ci_high": 5.987294132368352,
      "ci_metric": "octane_score",
      "median_latency_ms": 3924.2428545840085,
      "p95_latency_ms": 33934.06945816241,
      "input_tokens": 10929,
      "output_tokens": 4106,
      "reasoning_tokens": 33368,
      "undefined_reason": null
    },
    {
      "snapshot_id": "75a920ca-3b8e-42cd-8114-fcd2ed73a9cb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 79,
      "content_successes": 79,
      "success_rate": 0.8229166666666666,
      "total_cost_usd": 0.024997500000000002,
      "cost_per_success_usd": 0.00042120086145738323,
      "octane_score": 16.71367854139131,
      "ci_low": 7.282094446568545,
      "ci_high": 29.655453622583384,
      "ci_metric": "octane_score",
      "median_latency_ms": 711.5451665595174,
      "p95_latency_ms": 1506.6241039894521,
      "input_tokens": 10308,
      "output_tokens": 3837,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "81960822-21d5-4283-a95b-ee66b3d750b1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 94,
      "content_successes": 94,
      "success_rate": 0.9791666666666666,
      "total_cost_usd": 0.015002399999999999,
      "cost_per_success_usd": 0.0001635256866804693,
      "octane_score": 43.050213961257754,
      "ci_low": 23.1997901068527,
      "ci_high": 86.02146616810663,
      "ci_metric": "octane_score",
      "median_latency_ms": 1933.4572083316743,
      "p95_latency_ms": 7067.556052235886,
      "input_tokens": 10308,
      "output_tokens": 3443,
      "reasoning_tokens": 7341,
      "undefined_reason": null
    },
    {
      "snapshot_id": "def1896c-8917-4192-9002-f137f4653522",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 96,
      "content_successes": 96,
      "success_rate": 1.0,
      "total_cost_usd": 0.08876594,
      "cost_per_success_usd": 0.0009443685056590752,
      "octane_score": 7.454521998107836,
      "ci_low": 4.500660968464324,
      "ci_high": 13.004972247665528,
      "ci_metric": "octane_score",
      "median_latency_ms": 3426.100957673043,
      "p95_latency_ms": 14271.401624660939,
      "input_tokens": 212746,
      "output_tokens": 11739,
      "reasoning_tokens": 29278,
      "undefined_reason": null
    },
    {
      "snapshot_id": "66d9adfc-6b07-4e05-ab28-bb6842f89b2f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 93,
      "content_successes": 93,
      "success_rate": 0.96875,
      "total_cost_usd": 0.09790599999999999,
      "cost_per_success_usd": 0.0010956367786802569,
      "octane_score": 6.425318989597603,
      "ci_low": 3.4755938644626445,
      "ci_high": 12.136000663866962,
      "ci_metric": "octane_score",
      "median_latency_ms": 1614.8845213465393,
      "p95_latency_ms": 9198.579364456236,
      "input_tokens": 10308,
      "output_tokens": 3358,
      "reasoning_tokens": 4371,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4c6a6525-68dc-4ac6-88d7-7ad7f04b3d75",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 93,
      "content_successes": 93,
      "success_rate": 0.96875,
      "total_cost_usd": 1.706933,
      "cost_per_success_usd": 0.019206325713754843,
      "octane_score": 0.3665363122897656,
      "ci_low": 0.20838494229584004,
      "ci_high": 0.6371732108146573,
      "ci_metric": "octane_score",
      "median_latency_ms": 3306.0586457140744,
      "p95_latency_ms": 18581.238094018772,
      "input_tokens": 207151,
      "output_tokens": 11562,
      "reasoning_tokens": 18445,
      "undefined_reason": null
    },
    {
      "snapshot_id": "822ccd5d-f9e5-4f89-93eb-8db60e95677f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 93,
      "content_successes": 93,
      "success_rate": 0.96875,
      "total_cost_usd": 0.12264,
      "cost_per_success_usd": 0.001362172919254658,
      "octane_score": 5.168077929201371,
      "ci_low": 2.7013532398016586,
      "ci_high": 10.184971990402138,
      "ci_metric": "octane_score",
      "median_latency_ms": 1351.0013539344072,
      "p95_latency_ms": 6834.218385163695,
      "input_tokens": 10308,
      "output_tokens": 3379,
      "reasoning_tokens": 5123,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6f6513a3-4fbe-4747-87ef-c420a9ae34c9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 93,
      "content_successes": 93,
      "success_rate": 0.96875,
      "total_cost_usd": 0.3746821,
      "cost_per_success_usd": 0.0041535341372830065,
      "octane_score": 1.6948977827255385,
      "ci_low": 1.028604310835594,
      "ci_high": 2.908213940058645,
      "ci_metric": "octane_score",
      "median_latency_ms": 2727.2453741170466,
      "p95_latency_ms": 15398.080833954737,
      "input_tokens": 207508,
      "output_tokens": 11944,
      "reasoning_tokens": 21647,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9b27d976-6ced-43dc-85aa-304884841d1b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 92,
      "content_successes": 92,
      "success_rate": 0.9583333333333334,
      "total_cost_usd": 0.00564743,
      "cost_per_success_usd": 7.0398157997558e-05,
      "octane_score": 100.0,
      "ci_low": 3.191207310779499e-05,
      "ci_high": 0.00012236990153286428,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 4457.2092718444765,
      "p95_latency_ms": 30641.355219762772,
      "input_tokens": 15845,
      "output_tokens": 8444,
      "reasoning_tokens": 21980,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4941dfa4-8479-4aaa-9661-7900cb93c31c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 95,
      "content_successes": 95,
      "success_rate": 0.9895833333333334,
      "total_cost_usd": 0.2522492,
      "cost_per_success_usd": 0.0028250938651943865,
      "octane_score": 2.4918873976144544,
      "ci_low": 1.7538175097668198,
      "ci_high": 3.7587983649651004,
      "ci_metric": "octane_score",
      "median_latency_ms": 3571.7468336224556,
      "p95_latency_ms": 18023.96046835929,
      "input_tokens": 29652,
      "output_tokens": 3759,
      "reasoning_tokens": 33222,
      "undefined_reason": null
    },
    {
      "snapshot_id": "edf6d8ce-2076-41f8-a83b-c94eb392671d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 94,
      "content_successes": 94,
      "success_rate": 0.9791666666666666,
      "total_cost_usd": 0.371208,
      "cost_per_success_usd": 0.004390490103519668,
      "octane_score": 1.6034236802201833,
      "ci_low": 1.1814254329149405,
      "ci_high": 2.265517018073985,
      "ci_metric": "octane_score",
      "median_latency_ms": 4701.884312089533,
      "p95_latency_ms": 27191.605092957616,
      "input_tokens": 29652,
      "output_tokens": 3209,
      "reasoning_tokens": 53351,
      "undefined_reason": null
    },
    {
      "snapshot_id": "13c6fcb9-57fe-4e26-a9c5-4b14208165bc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 93,
      "content_successes": 93,
      "success_rate": 0.96875,
      "total_cost_usd": 0.29334239,
      "cost_per_success_usd": 0.0035218931831891135,
      "octane_score": 1.99887260447268,
      "ci_low": 1.204030689422031,
      "ci_high": 3.854482474366271,
      "ci_metric": "octane_score",
      "median_latency_ms": 11005.286895669997,
      "p95_latency_ms": 81303.45869809389,
      "input_tokens": 10560,
      "output_tokens": 7908,
      "reasoning_tokens": 120945,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a30c25c9-5165-42d4-a495-234be8c9996a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 94,
      "content_successes": 96,
      "success_rate": 0.9791666666666666,
      "total_cost_usd": 0.5526162,
      "cost_per_success_usd": 0.006340089267080745,
      "octane_score": 1.1103654070469011,
      "ci_low": 0.694969027413777,
      "ci_high": 2.1435880645223384,
      "ci_metric": "octane_score",
      "median_latency_ms": 2258.946062065661,
      "p95_latency_ms": 10711.063239956275,
      "input_tokens": 17745,
      "output_tokens": 7961,
      "reasoning_tokens": 27301,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c8b36aea-bdd7-4ebd-92e4-14bb97cca534",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 56,
      "content_successes": 81,
      "success_rate": 0.5833333333333334,
      "total_cost_usd": 0.013665,
      "cost_per_success_usd": 0.00033581249501415017,
      "octane_score": 20.96353144768828,
      "ci_low": 5.416110959393577,
      "ci_high": 47.85352979756143,
      "ci_metric": "octane_score",
      "median_latency_ms": 2186.2748959101737,
      "p95_latency_ms": 12107.015843736008,
      "input_tokens": 10689,
      "output_tokens": 14409,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6230b4ed-aaef-4934-96df-82f681efd342",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.13900015503168,
      "p95_latency_ms": 39.885333040729165,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "7730b8e6-5b38-456e-bd9f-76c553856b58",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 77,
      "content_successes": 95,
      "success_rate": 0.8020833333333334,
      "total_cost_usd": 1.04066,
      "cost_per_success_usd": 0.014922395226045885,
      "octane_score": 0.47176178442642686,
      "ci_low": 0.20242115831285273,
      "ci_high": 0.8557268945301536,
      "ci_metric": "octane_score",
      "median_latency_ms": 2819.1913329064846,
      "p95_latency_ms": 5119.110968662426,
      "input_tokens": 15771,
      "output_tokens": 8816,
      "reasoning_tokens": 8843,
      "undefined_reason": null
    },
    {
      "snapshot_id": "976ec75c-6a27-4c1f-99a5-75b402a524ed",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 53,
      "content_successes": 53,
      "success_rate": 0.5520833333333334,
      "total_cost_usd": 0.05366946,
      "cost_per_success_usd": 0.003021956831798942,
      "octane_score": 2.329555381360317,
      "ci_low": 1.1921158974213384,
      "ci_high": 17.434830047230022,
      "ci_metric": "octane_score",
      "median_latency_ms": 5187.854520510882,
      "p95_latency_ms": 34284.843291155994,
      "input_tokens": 11520,
      "output_tokens": 203541,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "abbeca9c-06c7-42de-941f-da7a96be0b3d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 92,
      "content_successes": 92,
      "success_rate": 0.9583333333333334,
      "total_cost_usd": 0.5942976,
      "cost_per_success_usd": 0.00792847014202172,
      "octane_score": 0.8879160384856645,
      "ci_low": 0.23829716526055317,
      "ci_high": 3.6282851510352416,
      "ci_metric": "octane_score",
      "median_latency_ms": 3214.5407926291227,
      "p95_latency_ms": 39318.04952141829,
      "input_tokens": 15921,
      "output_tokens": 4612,
      "reasoning_tokens": 89307,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9540db28-7eca-4b97-b4ba-279006a1ba91",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 90,
      "content_successes": 90,
      "success_rate": 0.9375,
      "total_cost_usd": 0.37740080000000004,
      "cost_per_success_usd": 0.004695625621151605,
      "octane_score": 1.499228509199011,
      "ci_low": 0.4728269650820332,
      "ci_high": 7.973002481744459,
      "ci_metric": "octane_score",
      "median_latency_ms": 4270.574792288244,
      "p95_latency_ms": 106480.4776776582,
      "input_tokens": 15472,
      "output_tokens": 30659,
      "reasoning_tokens": 85103,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c481abc9-a5f5-41e2-97de-f0e386158a00",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 90,
      "content_successes": 90,
      "success_rate": 0.9375,
      "total_cost_usd": 0.666678,
      "cost_per_success_usd": 0.009586456420538528,
      "octane_score": 0.7343501593219919,
      "ci_low": 0.22959857586460033,
      "ci_high": 3.375285716492178,
      "ci_metric": "octane_score",
      "median_latency_ms": 4166.432833299041,
      "p95_latency_ms": 89379.67143836431,
      "input_tokens": 15168,
      "output_tokens": 3321,
      "reasoning_tokens": 102736,
      "undefined_reason": null
    },
    {
      "snapshot_id": "24a388b3-ed6f-49e4-985e-e9b6bb5cf461",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 85,
      "content_successes": 86,
      "success_rate": 0.8854166666666666,
      "total_cost_usd": 0.9408645,
      "cost_per_success_usd": 0.014560653110807115,
      "octane_score": 0.4834821450784205,
      "ci_low": 0.23714611251052622,
      "ci_high": 1.3507198184795706,
      "ci_metric": "octane_score",
      "median_latency_ms": 10851.659187115729,
      "p95_latency_ms": 225741.400384577,
      "input_tokens": 13731,
      "output_tokens": 3712,
      "reasoning_tokens": 307621,
      "undefined_reason": null
    },
    {
      "snapshot_id": "be19d5aa-747d-4912-b578-63dcd82c4ec7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "overall",
      "vertical": null,
      "task_id": null,
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 96,
      "successes": 85,
      "content_successes": 85,
      "success_rate": 0.8854166666666666,
      "total_cost_usd": 0.716983,
      "cost_per_success_usd": 0.011891588320820124,
      "octane_score": 0.5919996227442801,
      "ci_low": 0.25481014429416693,
      "ci_high": 2.032882245178965,
      "ci_metric": "octane_score",
      "median_latency_ms": 12472.712728660554,
      "p95_latency_ms": 203136.58874016255,
      "input_tokens": 15171,
      "output_tokens": 3068,
      "reasoning_tokens": 280691,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5896b695-634e-4bad-9115-e1f5f0d9b00a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000414,
      "cost_per_success_usd": 0.000138,
      "octane_score": 22.021739130434785,
      "ci_low": 20.173913043478265,
      "ci_high": 23.6231884057971,
      "ci_metric": "octane_score",
      "median_latency_ms": 589.6747084334493,
      "p95_latency_ms": 2735.835070163011,
      "input_tokens": 294,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ea5a4842-e31d-455c-a345-7f7567f5cdc7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.000534,
      "cost_per_success_usd": 0.000178,
      "octane_score": 54.949438202247194,
      "ci_low": 42.00561797752809,
      "ci_high": 72.02247191011236,
      "ci_metric": "octane_score",
      "median_latency_ms": 611.0329581424594,
      "p95_latency_ms": 722.7195832878351,
      "input_tokens": 189,
      "output_tokens": 69,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7968b433-a39d-45c6-8f63-89b5dd07eba0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000486,
      "cost_per_success_usd": 0.000162,
      "octane_score": 14.45267489711934,
      "ci_low": 13.172839506172838,
      "ci_high": 15.617283950617285,
      "ci_metric": "octane_score",
      "median_latency_ms": 490.5822090804577,
      "p95_latency_ms": 871.0200461558996,
      "input_tokens": 411,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f4fcc3aa-76da-48bb-a829-4e350c94fd0e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000264,
      "cost_per_success_usd": 8.800000000000001e-05,
      "octane_score": 18.973484848484848,
      "ci_low": 18.909090909090907,
      "ci_high": 19.10227272727273,
      "ci_metric": "octane_score",
      "median_latency_ms": 506.55991677194834,
      "p95_latency_ms": 622.4372043274343,
      "input_tokens": 204,
      "output_tokens": 12,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6f505f51-98b1-4534-80fa-37935d87e30e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00093,
      "cost_per_success_usd": 0.00031,
      "octane_score": 9.499999999999998,
      "ci_low": 9.370967741935482,
      "ci_high": 9.648387096774194,
      "ci_metric": "octane_score",
      "median_latency_ms": 675.4067922011018,
      "p95_latency_ms": 737.0895294472575,
      "input_tokens": 360,
      "output_tokens": 114,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "15b84a36-1a74-481a-91d0-4cc8637ccf6b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0010170000000000001,
      "cost_per_success_usd": 0.00033900000000000005,
      "octane_score": 9.128810226155357,
      "ci_low": 7.056047197640117,
      "ci_high": 10.51622418879056,
      "ci_metric": "octane_score",
      "median_latency_ms": 876.4199996367097,
      "p95_latency_ms": 923.3900618739426,
      "input_tokens": 327,
      "output_tokens": 138,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c0c65b45-7723-464e-bee4-6d98d392e964",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.005084999999999999,
      "cost_per_success_usd": 0.0016949999999999997,
      "octane_score": 1.825762045231072,
      "ci_low": 1.411209439528024,
      "ci_high": 2.1032448377581123,
      "ci_metric": "octane_score",
      "median_latency_ms": 2267.505374737084,
      "p95_latency_ms": 2373.0853509157896,
      "input_tokens": 327,
      "output_tokens": 138,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8fbfcc34-0b32-4c5c-9def-1d6e7833d8d8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004275,
      "cost_per_success_usd": 0.001425,
      "octane_score": 2.0666666666666664,
      "ci_low": 2.03859649122807,
      "ci_high": 2.0989473684210527,
      "ci_metric": "octane_score",
      "median_latency_ms": 1816.44191686064,
      "p95_latency_ms": 1863.2016805000603,
      "input_tokens": 360,
      "output_tokens": 99,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cacd647d-1720-4762-b508-d941471dd8c5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.016545,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 6006.054666824639,
      "p95_latency_ms": 6923.1119537726045,
      "input_tokens": 204,
      "output_tokens": 621,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "b909562a-d7b6-460d-9626-a1d266b8640f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00243,
      "cost_per_success_usd": 0.00081,
      "octane_score": 2.8905349794238684,
      "ci_low": 2.634567901234568,
      "ci_high": 3.1234567901234573,
      "ci_metric": "octane_score",
      "median_latency_ms": 1482.2169998660684,
      "p95_latency_ms": 1541.3209373131394,
      "input_tokens": 411,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "be5cf5c3-54f5-434c-aeeb-99cbac183476",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.01722,
      "cost_per_success_usd": 0.0057399999999999994,
      "octane_score": 1.704006968641115,
      "ci_low": 1.3026132404181185,
      "ci_high": 2.2334494773519165,
      "ci_metric": "octane_score",
      "median_latency_ms": 4041.6594995185733,
      "p95_latency_ms": 4067.3022248782218,
      "input_tokens": 189,
      "output_tokens": 651,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6b2aff79-c3cc-4988-99ff-ee30506ee993",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0073950000000000005,
      "cost_per_success_usd": 0.002465,
      "octane_score": 1.2328600405679513,
      "ci_low": 1.1294117647058826,
      "ci_high": 1.3225152129817443,
      "ci_metric": "octane_score",
      "median_latency_ms": 3027.725833468139,
      "p95_latency_ms": 3648.0097712017596,
      "input_tokens": 294,
      "output_tokens": 237,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ea3ae200-d6ab-41fb-964c-e4a8d6d73c4a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001242,
      "cost_per_success_usd": 0.00041400000000000003,
      "octane_score": 7.340579710144927,
      "ci_low": 6.724637681159421,
      "ci_high": 7.8743961352657,
      "ci_metric": "octane_score",
      "median_latency_ms": 1188.1498750299215,
      "p95_latency_ms": 1566.437199898064,
      "input_tokens": 294,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "955dfb27-4285-49a0-a1be-11c46bca4b0a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.006507,
      "cost_per_success_usd": 0.002169,
      "octane_score": 4.509451360073767,
      "ci_low": 3.447210696173352,
      "ci_high": 5.91055786076533,
      "ci_metric": "octane_score",
      "median_latency_ms": 2875.897583551705,
      "p95_latency_ms": 3401.0874459519982,
      "input_tokens": 189,
      "output_tokens": 396,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "83931eaa-6b24-45f6-801a-153c2df96edd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001458,
      "cost_per_success_usd": 0.00048600000000000005,
      "octane_score": 4.81755829903978,
      "ci_low": 4.390946502057613,
      "ci_high": 5.205761316872428,
      "ci_metric": "octane_score",
      "median_latency_ms": 1643.9853329211473,
      "p95_latency_ms": 1768.8629580661654,
      "input_tokens": 411,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "81edf534-2ee7-47ff-a21c-4b30a31c51f9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0007920000000000001,
      "cost_per_success_usd": 0.000264,
      "octane_score": 6.32449494949495,
      "ci_low": 6.303030303030302,
      "ci_high": 6.367424242424243,
      "ci_metric": "octane_score",
      "median_latency_ms": 1130.0594583153725,
      "p95_latency_ms": 1470.4946951940653,
      "input_tokens": 204,
      "output_tokens": 12,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "af6a9a72-7a6a-4f88-9d4b-b0256e065c4b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00279,
      "cost_per_success_usd": 0.0009299999999999998,
      "octane_score": 3.166666666666667,
      "ci_low": 3.123655913978495,
      "ci_high": 3.216129032258065,
      "ci_metric": "octane_score",
      "median_latency_ms": 1288.1288332864642,
      "p95_latency_ms": 1310.9274079091847,
      "input_tokens": 360,
      "output_tokens": 114,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5afb3bc3-264b-42d6-961f-9dda08046f0a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003051,
      "cost_per_success_usd": 0.001017,
      "octane_score": 3.0429367420517863,
      "ci_low": 2.352015732546706,
      "ci_high": 3.5054080629301874,
      "ci_metric": "octane_score",
      "median_latency_ms": 1638.7416245415807,
      "p95_latency_ms": 3108.493750263005,
      "input_tokens": 327,
      "output_tokens": 138,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6b76ad75-3c3a-4c9d-8b7b-418dd8f04ac1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 5.298e-05,
      "cost_per_success_usd": 1.766e-05,
      "octane_score": 94.54511136277843,
      "ci_low": 94.2242355605889,
      "ci_high": 95.18686296715744,
      "ci_metric": "octane_score",
      "median_latency_ms": 754.5811254531145,
      "p95_latency_ms": 1166.4052629843354,
      "input_tokens": 195,
      "output_tokens": 6,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5f5bba8e-33b7-452b-abdf-23d8dff3d77b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 7.59e-05,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 1544.8329579085112,
      "p95_latency_ms": 1617.456096317619,
      "input_tokens": 270,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "f0e24404-a65c-4183-bfc3-67b4009c2c10",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00014159999999999997,
      "cost_per_success_usd": 4.719999999999999e-05,
      "octane_score": 65.56497175141244,
      "ci_low": 50.67796610169493,
      "ci_high": 75.52966101694918,
      "ci_metric": "octane_score",
      "median_latency_ms": 2293.0038748309016,
      "p95_latency_ms": 2446.178700309247,
      "input_tokens": 321,
      "output_tokens": 153,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a6aed3f6-e4e6-4da3-aaa0-7c511498d672",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 6.198e-05,
      "cost_per_success_usd": 2.066e-05,
      "octane_score": 473.4269119070668,
      "ci_low": 361.9070667957406,
      "ci_high": 620.5227492739593,
      "ci_metric": "octane_score",
      "median_latency_ms": 902.4023748934269,
      "p95_latency_ms": 1631.9008999504147,
      "input_tokens": 177,
      "output_tokens": 42,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "37796e94-c475-46a4-8696-73155bf980e0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00010674,
      "cost_per_success_usd": 3.558e-05,
      "octane_score": 65.80475922803072,
      "ci_low": 59.97751545812254,
      "ci_high": 71.10736368746487,
      "ci_metric": "octane_score",
      "median_latency_ms": 959.9609998986125,
      "p95_latency_ms": 1229.328038636595,
      "input_tokens": 393,
      "output_tokens": 12,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8d99d5f7-d88c-47fc-b769-5fb9d9c9602c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00011738,
      "cost_per_success_usd": 3.912666666666667e-05,
      "octane_score": 75.26835917532797,
      "ci_low": 69.6022275258552,
      "ci_high": 87.36351002580899,
      "ci_metric": "octane_score",
      "median_latency_ms": 1211.2876251339912,
      "p95_latency_ms": 1326.176261343062,
      "input_tokens": 330,
      "output_tokens": 105,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ddf958ec-2c67-4d6e-8360-bfe7b9f6c85e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00010513999999999999,
      "cost_per_success_usd": 3.5046666666666664e-05,
      "octane_score": 86.71295415636295,
      "ci_low": 69.86782666889746,
      "ci_high": 105.71963852766147,
      "ci_metric": "octane_score",
      "median_latency_ms": 1791.7446251958609,
      "p95_latency_ms": 44198.147125076495,
      "input_tokens": 331,
      "output_tokens": 64,
      "reasoning_tokens": 373,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6d0a1e57-b624-40d0-a834-48b7b30376ba",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00018432,
      "cost_per_success_usd": 6.144e-05,
      "octane_score": 50.368923611111114,
      "ci_low": 35.67783935954957,
      "ci_high": 80.10938399539437,
      "ci_metric": "octane_score",
      "median_latency_ms": 6056.837541982532,
      "p95_latency_ms": 6785.132016707212,
      "input_tokens": 540,
      "output_tokens": 402,
      "reasoning_tokens": 382,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0ed77ffb-b840-4676-af42-df73b4f19ff6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 9.194e-05,
      "cost_per_success_usd": 3.064666666666667e-05,
      "octane_score": 96.09527953012832,
      "ci_low": 84.68648751667618,
      "ci_high": 104.85425481899388,
      "ci_metric": "octane_score",
      "median_latency_ms": 2864.914333447814,
      "p95_latency_ms": 5219.251408334821,
      "input_tokens": 391,
      "output_tokens": 181,
      "reasoning_tokens": 156,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2a0075b0-674c-4d1b-89f6-3f94e8dd076c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 9.510000000000001e-05,
      "cost_per_success_usd": 3.1700000000000005e-05,
      "octane_score": 52.67087276550999,
      "ci_low": 30.29148524431544,
      "ci_high": 86.42167011732232,
      "ci_metric": "octane_score",
      "median_latency_ms": 1538.4028749540448,
      "p95_latency_ms": 23921.23026233166,
      "input_tokens": 336,
      "output_tokens": 6,
      "reasoning_tokens": 373,
      "undefined_reason": null
    },
    {
      "snapshot_id": "06be008e-982d-420d-9062-05ad6a135ff9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00017203999999999999,
      "cost_per_success_usd": 5.7346666666666664e-05,
      "octane_score": 40.82771448500349,
      "ci_low": 29.308237594720122,
      "ci_high": 58.33077159545036,
      "ci_metric": "octane_score",
      "median_latency_ms": 5750.235499814153,
      "p95_latency_ms": 30410.553274862465,
      "input_tokens": 454,
      "output_tokens": 10,
      "reasoning_tokens": 744,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a37de241-d354-46ad-983a-f858306e9c37",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00013714,
      "cost_per_success_usd": 4.5713333333333335e-05,
      "octane_score": 213.9638325798454,
      "ci_low": 148.33127317676144,
      "ci_high": 314.7039897039897,
      "ci_metric": "octane_score",
      "median_latency_ms": 2303.5247921943665,
      "p95_latency_ms": 4862.8891413100055,
      "input_tokens": 317,
      "output_tokens": 166,
      "reasoning_tokens": 455,
      "undefined_reason": null
    },
    {
      "snapshot_id": "684c1c8e-de23-4ad2-83d3-f6db3ead80f2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00021924,
      "cost_per_success_usd": 7.308e-05,
      "octane_score": 22.84710819193578,
      "ci_low": 13.806885526061912,
      "ci_high": 60.925621845162084,
      "ci_metric": "octane_score",
      "median_latency_ms": 1330.732500180602,
      "p95_latency_ms": 2218.6009869910777,
      "input_tokens": 256,
      "output_tokens": 7,
      "reasoning_tokens": 117,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4048d403-4a6e-445a-a77b-75d52c2b38cd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00075168,
      "cost_per_success_usd": 0.00025056,
      "octane_score": 9.344401873137505,
      "ci_low": 8.521689938028748,
      "ci_high": 10.16535832203088,
      "ci_metric": "octane_score",
      "median_latency_ms": 3892.615457996726,
      "p95_latency_ms": 7140.590095892548,
      "input_tokens": 612,
      "output_tokens": 11,
      "reasoning_tokens": 547,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b279a832-eafa-4527-8adc-372ea512333e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00084129,
      "cost_per_success_usd": 0.00028043,
      "octane_score": 34.87857932460863,
      "ci_low": 23.24865520350735,
      "ci_high": 56.495119274533586,
      "ci_metric": "octane_score",
      "median_latency_ms": 4186.591208912432,
      "p95_latency_ms": 5259.01902122423,
      "input_tokens": 396,
      "output_tokens": 42,
      "reasoning_tokens": 727,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0ab64266-fe90-4b81-a648-bafb8158e3eb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000582465,
      "cost_per_success_usd": 0.000194155,
      "octane_score": 15.652442636038217,
      "ci_low": 13.107330012292167,
      "ci_high": 18.34010899554464,
      "ci_metric": "octane_score",
      "median_latency_ms": 2841.610665433109,
      "p95_latency_ms": 7482.229854259639,
      "input_tokens": 489,
      "output_tokens": 15,
      "reasoning_tokens": 410,
      "undefined_reason": null
    },
    {
      "snapshot_id": "989e333a-7c8f-421c-b97e-c06a7cee9e8e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00043239,
      "cost_per_success_usd": 0.000216195,
      "octane_score": 13.621961654987395,
      "ci_low": 8.661863834277627,
      "ci_high": 17.7009483010026,
      "ci_metric": "octane_score",
      "median_latency_ms": 2960.3649163618684,
      "p95_latency_ms": 5967.5103921443215,
      "input_tokens": 470,
      "output_tokens": 86,
      "reasoning_tokens": 176,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ce9dcaa3-5e8a-447d-9bd7-b2801bd1e61e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0008304149999999999,
      "cost_per_success_usd": 0.000276805,
      "octane_score": 11.17995219257841,
      "ci_low": 7.699324940704251,
      "ci_high": 17.048993233879514,
      "ci_metric": "octane_score",
      "median_latency_ms": 5089.218750596046,
      "p95_latency_ms": 6271.37835053727,
      "input_tokens": 461,
      "output_tokens": 117,
      "reasoning_tokens": 607,
      "undefined_reason": null
    },
    {
      "snapshot_id": "935ffe21-d401-4956-ac34-d5ec07698366",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00013769999999999999,
      "cost_per_success_usd": 4.59e-05,
      "octane_score": 51.00944081336238,
      "ci_low": 46.492374727668846,
      "ci_high": 55.11982570806101,
      "ci_metric": "octane_score",
      "median_latency_ms": 584.5807073637843,
      "p95_latency_ms": 713.6495950631796,
      "input_tokens": 384,
      "output_tokens": 9,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9f930de5-acb9-4af5-942f-ad974e87ac41",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005409,
      "cost_per_success_usd": 0.0001803,
      "octane_score": 17.163985949343687,
      "ci_low": 13.266777592900722,
      "ci_high": 19.772601220188577,
      "ci_metric": "octane_score",
      "median_latency_ms": 764.3545418977737,
      "p95_latency_ms": 802.4713044054806,
      "input_tokens": 303,
      "output_tokens": 180,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "67d9264c-ba41-430a-83d0-7ec75222ccfb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0004925999999999999,
      "cost_per_success_usd": 0.00016419999999999998,
      "octane_score": 17.935444579780757,
      "ci_low": 17.69183922046285,
      "ci_high": 18.215590742996348,
      "ci_metric": "octane_score",
      "median_latency_ms": 785.711582750082,
      "p95_latency_ms": 823.2944574207067,
      "input_tokens": 342,
      "output_tokens": 156,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5aa44127-3f7c-4bde-a126-836a98889bac",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 5.7900000000000005e-05,
      "cost_per_success_usd": 1.93e-05,
      "octane_score": 86.5112262521589,
      "ci_low": 86.21761658031087,
      "ci_high": 87.09844559585494,
      "ci_metric": "octane_score",
      "median_latency_ms": 523.8693747669458,
      "p95_latency_ms": 537.533549964428,
      "input_tokens": 168,
      "output_tokens": 3,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "09b5a398-ccf9-4d9d-b678-a3e4d7bb1d89",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00023520000000000002,
      "cost_per_success_usd": 7.840000000000001e-05,
      "octane_score": 124.75765306122447,
      "ci_low": 95.36989795918366,
      "ci_high": 163.52040816326527,
      "ci_metric": "octane_score",
      "median_latency_ms": 753.1672902405262,
      "p95_latency_ms": 809.2243161983788,
      "input_tokens": 159,
      "output_tokens": 75,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "16a648cd-dba5-4cb0-bf4c-5c63c11927df",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001335,
      "cost_per_success_usd": 4.45e-05,
      "octane_score": 68.29213483146067,
      "ci_low": 62.561797752809,
      "ci_high": 73.25842696629213,
      "ci_metric": "octane_score",
      "median_latency_ms": 705.1329584792256,
      "p95_latency_ms": 853.336483053863,
      "input_tokens": 270,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9cc2fa20-cea5-40cb-aa49-9a1e15b222a2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.012181500000000001,
      "cost_per_success_usd": 0.004060500000000001,
      "octane_score": 0.7484299963058735,
      "ci_low": 0.663594823945014,
      "ci_high": 0.8548523206751054,
      "ci_metric": "octane_score",
      "median_latency_ms": 2857.2678342461586,
      "p95_latency_ms": 3356.286333873868,
      "input_tokens": 271,
      "output_tokens": 21,
      "reasoning_tokens": 1549,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f6199995-cda1-4d0d-a51d-398f72daac63",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0070665,
      "cost_per_success_usd": 0.0023555,
      "octane_score": 0.7088374725819007,
      "ci_low": 0.6657363104731526,
      "ci_high": 0.7693018955154877,
      "ci_metric": "octane_score",
      "median_latency_ms": 2244.0388752147555,
      "p95_latency_ms": 2476.280425488949,
      "input_tokens": 171,
      "output_tokens": 3,
      "reasoning_tokens": 905,
      "undefined_reason": null
    },
    {
      "snapshot_id": "10ba57fb-8ac7-43b1-b25c-34fec29ef81b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0219195,
      "cost_per_success_usd": 0.0073065,
      "octane_score": 0.40306576336139055,
      "ci_low": 0.2963522012578616,
      "ci_high": 0.6758043461567993,
      "ci_metric": "octane_score",
      "median_latency_ms": 6123.089167289436,
      "p95_latency_ms": 7016.5343538858,
      "input_tokens": 343,
      "output_tokens": 117,
      "reasoning_tokens": 2737,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5bf1c49b-8b5a-4b5f-99fa-e18887f09a25",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0128805,
      "cost_per_success_usd": 0.0042935,
      "octane_score": 2.278094794456737,
      "ci_low": 1.6883820706785597,
      "ci_high": 3.08237803252466,
      "ci_metric": "octane_score",
      "median_latency_ms": 3288.6977503076196,
      "p95_latency_ms": 3421.4556619524956,
      "input_tokens": 162,
      "output_tokens": 67,
      "reasoning_tokens": 1618,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b9775069-1cd1-41b1-a095-bcd1d2ed058d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.008079,
      "cost_per_success_usd": 0.0026929999999999996,
      "octane_score": 0.8694145315014236,
      "ci_low": 0.7539100267303643,
      "ci_high": 1.0368353377202932,
      "ci_metric": "octane_score",
      "median_latency_ms": 2286.5978749468923,
      "p95_latency_ms": 2402.7015501633286,
      "input_tokens": 386,
      "output_tokens": 9,
      "reasoning_tokens": 991,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fdbaf981-4626-4598-8803-7a6fecede56f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.016437,
      "cost_per_success_usd": 0.005479,
      "octane_score": 0.564823264585995,
      "ci_low": 0.41505475386347357,
      "ci_high": 0.7843534353435344,
      "ci_metric": "octane_score",
      "median_latency_ms": 3640.7849164679646,
      "p95_latency_ms": 9906.062129046766,
      "input_tokens": 303,
      "output_tokens": 189,
      "reasoning_tokens": 1942,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0be658e6-978b-4c5b-8c53-0e449608d3ea",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002211375,
      "cost_per_success_usd": 0.0007371249999999999,
      "octane_score": 3.9952518229608276,
      "ci_low": 3.7825150732127475,
      "ci_high": 4.372896093410346,
      "ci_metric": "octane_score",
      "median_latency_ms": 3144.20691691339,
      "p95_latency_ms": 3989.4480667077005,
      "input_tokens": 342,
      "output_tokens": 132,
      "reasoning_tokens": 979,
      "undefined_reason": null
    },
    {
      "snapshot_id": "229b1374-c602-4a7c-aaee-e45794d9917d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001336125,
      "cost_per_success_usd": 0.000445375,
      "octane_score": 3.748900739077557,
      "ci_low": 3.283513602097673,
      "ci_high": 4.897718910963944,
      "ci_metric": "octane_score",
      "median_latency_ms": 2752.7259597554803,
      "p95_latency_ms": 3315.1086206547916,
      "input_tokens": 168,
      "output_tokens": 3,
      "reasoning_tokens": 676,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e08b5941-3061-42dc-970b-8913f7d124d7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00149775,
      "cost_per_success_usd": 0.00049925,
      "octane_score": 4.689701218494408,
      "ci_low": 3.9285767207527598,
      "ci_high": 5.957591178965226,
      "ci_metric": "octane_score",
      "median_latency_ms": 2785.285457968712,
      "p95_latency_ms": 3221.7783336527646,
      "input_tokens": 384,
      "output_tokens": 9,
      "reasoning_tokens": 713,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f61b1eb6-1c7a-46ce-8da3-db27ae4b2e27",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001990875,
      "cost_per_success_usd": 0.000663625,
      "octane_score": 14.738745526464493,
      "ci_low": 10.004348553269779,
      "ci_high": 22.235862460926395,
      "ci_metric": "octane_score",
      "median_latency_ms": 3201.108332723379,
      "p95_latency_ms": 3262.052095402032,
      "input_tokens": 159,
      "output_tokens": 63,
      "reasoning_tokens": 967,
      "undefined_reason": null
    },
    {
      "snapshot_id": "54575534-dcbd-456c-b44f-715dba2c2772",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002015625,
      "cost_per_success_usd": 0.0006718750000000001,
      "octane_score": 4.523162790697674,
      "ci_low": 4.096474074074074,
      "ci_high": 4.991479674796748,
      "ci_metric": "octane_score",
      "median_latency_ms": 3093.2897087186575,
      "p95_latency_ms": 3441.036245878786,
      "input_tokens": 270,
      "output_tokens": 21,
      "reasoning_tokens": 1000,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ad20dd7c-e50e-4bdc-88c2-2cf7a5ed1ac9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00329925,
      "cost_per_success_usd": 0.00109975,
      "octane_score": 2.8139728726225655,
      "ci_low": 2.0613107150798946,
      "ci_high": 3.9258529259937114,
      "ci_metric": "octane_score",
      "median_latency_ms": 5167.343125678599,
      "p95_latency_ms": 7031.564462278038,
      "input_tokens": 303,
      "output_tokens": 161,
      "reasoning_tokens": 1538,
      "undefined_reason": null
    },
    {
      "snapshot_id": "20ff232c-0357-452c-b28e-c7b551fa0fce",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00077072,
      "cost_per_success_usd": 0.0002569066666666667,
      "octane_score": 11.829198671372222,
      "ci_low": 10.836620303093213,
      "ci_high": 12.682850918145036,
      "ci_metric": "octane_score",
      "median_latency_ms": 10263.839417137206,
      "p95_latency_ms": 10892.187054175884,
      "input_tokens": 252,
      "output_tokens": 481,
      "reasoning_tokens": 1408,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b4972c84-b494-4dd9-a351-b040b5266914",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00101016,
      "cost_per_success_usd": 0.00033672,
      "octane_score": 29.047873604181515,
      "ci_low": 21.410219811590064,
      "ci_high": 38.56742572495997,
      "ci_metric": "octane_score",
      "median_latency_ms": 19465.62649961561,
      "p95_latency_ms": 30395.078975055367,
      "input_tokens": 156,
      "output_tokens": 763,
      "reasoning_tokens": 1739,
      "undefined_reason": null
    },
    {
      "snapshot_id": "69a02ac7-c3fb-4411-a432-3f993440cf0c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005369000000000001,
      "cost_per_success_usd": 0.0001789666666666667,
      "octane_score": 13.082510709629352,
      "ci_low": 11.924008195194634,
      "ci_high": 14.157603510780385,
      "ci_metric": "octane_score",
      "median_latency_ms": 7675.349459052086,
      "p95_latency_ms": 11918.024945352225,
      "input_tokens": 375,
      "output_tokens": 11,
      "reasoning_tokens": 1275,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c4b10d87-1da2-45d0-a42b-da0b8a743744",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00092324,
      "cost_per_success_usd": 0.00030774666666666666,
      "octane_score": 5.425458169056801,
      "ci_low": 3.58389857187831,
      "ci_high": 7.663239703812497,
      "ci_metric": "octane_score",
      "median_latency_ms": 9198.831042274833,
      "p95_latency_ms": 18721.92484131083,
      "input_tokens": 174,
      "output_tokens": 0,
      "reasoning_tokens": 2282,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2c850917-6226-4666-82cb-2f99ba62497e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00075236,
      "cost_per_success_usd": 0.00025078666666666666,
      "octane_score": 11.74304854059227,
      "ci_low": 11.41630565269586,
      "ci_high": 12.103402926187732,
      "ci_metric": "octane_score",
      "median_latency_ms": 12729.858040809631,
      "p95_latency_ms": 13222.884716838598,
      "input_tokens": 306,
      "output_tokens": 494,
      "reasoning_tokens": 1349,
      "undefined_reason": null
    },
    {
      "snapshot_id": "816b20bd-ae3e-479d-b342-1125abf73bae",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0018604800000000003,
      "cost_per_success_usd": 0.0006201600000000001,
      "octane_score": 4.990110079119366,
      "ci_low": 3.5937403078449957,
      "ci_high": 7.15619264603506,
      "ci_metric": "octane_score",
      "median_latency_ms": 29693.386582657695,
      "p95_latency_ms": 59365.62045803293,
      "input_tokens": 288,
      "output_tokens": 184,
      "reasoning_tokens": 4436,
      "undefined_reason": null
    },
    {
      "snapshot_id": "485a18be-05e9-48e3-84f0-195bb6707cf3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00346716,
      "cost_per_success_usd": 0.0011557199999999998,
      "octane_score": 2.5481950645485067,
      "ci_low": 2.4238142730496453,
      "ci_high": 2.763647097244202,
      "ci_metric": "octane_score",
      "median_latency_ms": 8652.90195774287,
      "p95_latency_ms": 22559.238682873543,
      "input_tokens": 313,
      "output_tokens": 205,
      "reasoning_tokens": 1503,
      "undefined_reason": null
    },
    {
      "snapshot_id": "aed714cc-a5f5-447b-81db-b01c3e774aa7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004430519999999999,
      "cost_per_success_usd": 0.0014768399999999999,
      "octane_score": 2.095465092133655,
      "ci_low": 1.5133027351766857,
      "ci_high": 3.0546320732083494,
      "ci_metric": "octane_score",
      "median_latency_ms": 7190.12691732496,
      "p95_latency_ms": 9344.473866280168,
      "input_tokens": 309,
      "output_tokens": 119,
      "reasoning_tokens": 2124,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a2429438-e3b0-425c-991f-ac2ba6f963e8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00392088,
      "cost_per_success_usd": 0.00130696,
      "octane_score": 1.2775193323947687,
      "ci_low": 1.1429184235985617,
      "ci_high": 1.416604448064436,
      "ci_metric": "octane_score",
      "median_latency_ms": 95716.72466769814,
      "p95_latency_ms": 109541.65286738425,
      "input_tokens": 174,
      "output_tokens": 9,
      "reasoning_tokens": 2008,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1af8596a-ec11-4708-bd15-2bc0b8b644c2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.005576039999999999,
      "cost_per_success_usd": 0.0018586799999999997,
      "octane_score": 1.2596753251411397,
      "ci_low": 1.151298042685427,
      "ci_high": 1.3574417855993135,
      "ci_metric": "octane_score",
      "median_latency_ms": 26971.595959737897,
      "p95_latency_ms": 28698.109870869666,
      "input_tokens": 375,
      "output_tokens": 0,
      "reasoning_tokens": 2835,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4efe1feb-5507-4cc3-8d07-aab9c03f6622",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0031313999999999995,
      "cost_per_success_usd": 0.0010438,
      "octane_score": 9.370569074535352,
      "ci_low": 6.891798759476225,
      "ci_high": 12.475183426845058,
      "ci_metric": "octane_score",
      "median_latency_ms": 10675.538665615022,
      "p95_latency_ms": 16610.275840945542,
      "input_tokens": 163,
      "output_tokens": 190,
      "reasoning_tokens": 1390,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b7387383-a230-4379-8c2e-1f2e9efc3301",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00292416,
      "cost_per_success_usd": 0.00097472,
      "octane_score": 3.1178184504267894,
      "ci_low": 2.7862688954238255,
      "ci_high": 3.4921224299609763,
      "ci_metric": "octane_score",
      "median_latency_ms": 10653.580417856574,
      "p95_latency_ms": 62317.692853976034,
      "input_tokens": 252,
      "output_tokens": 18,
      "reasoning_tokens": 1427,
      "undefined_reason": null
    },
    {
      "snapshot_id": "28fcf947-bd25-4d8a-b950-e2fe859ddc85",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0031537692,
      "cost_per_success_usd": 0.0010512564,
      "octane_score": 1.5882582656968054,
      "ci_low": 1.2980403193246566,
      "ci_high": 1.81077804692788,
      "ci_metric": "octane_score",
      "median_latency_ms": 3878.621750511229,
      "p95_latency_ms": 10819.201287440954,
      "input_tokens": 195,
      "output_tokens": 6,
      "reasoning_tokens": 985,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d9c2e3cd-1c69-4485-becb-feee78aec11d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002610822,
      "cost_per_success_usd": 0.000870274,
      "octane_score": 3.3839917083585167,
      "ci_low": 3.176502897229148,
      "ci_high": 3.7744494347289135,
      "ci_metric": "octane_score",
      "median_latency_ms": 3078.906917013228,
      "p95_latency_ms": 3190.618441440165,
      "input_tokens": 327,
      "output_tokens": 160,
      "reasoning_tokens": 643,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4b7a5259-b826-4cd5-856f-973902f459fd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003542598,
      "cost_per_success_usd": 0.001180866,
      "octane_score": 8.282904241463468,
      "ci_low": 6.334557489180336,
      "ci_high": 10.851704767267332,
      "ci_metric": "octane_score",
      "median_latency_ms": 3817.323957569897,
      "p95_latency_ms": 4362.35462045297,
      "input_tokens": 177,
      "output_tokens": 44,
      "reasoning_tokens": 1106,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2da6210a-5532-43d1-b835-6e6636b7ee0a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0069182988,
      "cost_per_success_usd": 0.0023060996,
      "octane_score": 1.3419483992220747,
      "ci_low": 0.8117103113963463,
      "ci_high": 2.3795251345415904,
      "ci_metric": "octane_score",
      "median_latency_ms": 5833.2232078537345,
      "p95_latency_ms": 11703.576858434824,
      "input_tokens": 309,
      "output_tokens": 136,
      "reasoning_tokens": 2090,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ac2231c6-ed0a-4315-83ed-ff6413d2b707",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0015180828,
      "cost_per_success_usd": 0.0005060276,
      "octane_score": 6.005601275503549,
      "ci_low": 5.501676193156263,
      "ci_high": 6.432512681521167,
      "ci_metric": "octane_score",
      "median_latency_ms": 2404.1475830599666,
      "p95_latency_ms": 2441.743845306337,
      "input_tokens": 273,
      "output_tokens": 18,
      "reasoning_tokens": 456,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6e0e37c5-25a4-4504-9160-55a6f997ef9b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0015906984,
      "cost_per_success_usd": 0.0005302328,
      "octane_score": 4.415670500454391,
      "ci_low": 4.040069003015474,
      "ci_high": 4.764544856024118,
      "ci_metric": "octane_score",
      "median_latency_ms": 2096.5452073141932,
      "p95_latency_ms": 2245.3687580302358,
      "input_tokens": 396,
      "output_tokens": 15,
      "reasoning_tokens": 470,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c4cb8c08-6623-4dfd-a9f9-37e69ac8772f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0033930800000000006,
      "cost_per_success_usd": 0.0011310266666666668,
      "octane_score": 2.736157119784974,
      "ci_low": 2.1231523012651334,
      "ci_high": 3.1572686320213523,
      "ci_metric": "octane_score",
      "median_latency_ms": 4123.095833696425,
      "p95_latency_ms": 4537.119007948786,
      "input_tokens": 309,
      "output_tokens": 120,
      "reasoning_tokens": 586,
      "undefined_reason": null
    },
    {
      "snapshot_id": "753ccb5f-afa3-4719-bb98-9f330685ee6f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00133188,
      "cost_per_success_usd": 0.00044395999999999997,
      "octane_score": 6.845211280295523,
      "ci_low": 6.279510390664944,
      "ci_high": 7.332873466694658,
      "ci_metric": "octane_score",
      "median_latency_ms": 1902.5924997404218,
      "p95_latency_ms": 2820.1223248615856,
      "input_tokens": 273,
      "output_tokens": 18,
      "reasoning_tokens": 231,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fefda3c0-0535-444b-a786-dcac2bc1d619",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0018758,
      "cost_per_success_usd": 0.0006252666666666666,
      "octane_score": 15.64292568504105,
      "ci_low": 11.95809787823862,
      "ci_high": 20.503251945836446,
      "ci_metric": "octane_score",
      "median_latency_ms": 3646.8753749504685,
      "p95_latency_ms": 4043.0688380263746,
      "input_tokens": 177,
      "output_tokens": 42,
      "reasoning_tokens": 328,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ce7dd771-dec8-4df8-aadd-ffa2abfebe21",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00242296,
      "cost_per_success_usd": 0.0008076533333333333,
      "octane_score": 2.8989335358404595,
      "ci_low": 2.3324230736747276,
      "ci_high": 3.888741252546726,
      "ci_metric": "octane_score",
      "median_latency_ms": 4159.810249693692,
      "p95_latency_ms": 4480.366586986929,
      "input_tokens": 396,
      "output_tokens": 15,
      "reasoning_tokens": 476,
      "undefined_reason": null
    },
    {
      "snapshot_id": "06677917-789f-4253-88d5-2390f6784f02",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0060766,
      "cost_per_success_usd": 0.002025533333333333,
      "octane_score": 0.8243096468419843,
      "ci_low": 0.7943984521204048,
      "ci_high": 0.858174316658071,
      "ci_metric": "octane_score",
      "median_latency_ms": 10779.978208243847,
      "p95_latency_ms": 13312.232195120305,
      "input_tokens": 195,
      "output_tokens": 9,
      "reasoning_tokens": 1310,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b8f0ac03-1821-4176-90ca-8adf20c169a4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0023534800000000002,
      "cost_per_success_usd": 0.0007844933333333334,
      "octane_score": 3.754015330489317,
      "ci_low": 3.6927453529401832,
      "ci_high": 3.8161114248070764,
      "ci_metric": "octane_score",
      "median_latency_ms": 3288.5247096419334,
      "p95_latency_ms": 3298.051657900214,
      "input_tokens": 327,
      "output_tokens": 93,
      "reasoning_tokens": 371,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3e9735e6-0f02-4e4d-bc79-e274a7ae6b7f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0007019999999999999,
      "cost_per_success_usd": 0.00023399999999999997,
      "octane_score": 13.225071225071227,
      "ci_low": 10.222222222222223,
      "ci_high": 15.235042735042738,
      "ci_metric": "octane_score",
      "median_latency_ms": 789.8147497326136,
      "p95_latency_ms": 812.6759874634445,
      "input_tokens": 288,
      "output_tokens": 108,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5d06dc6b-5745-4e71-b22f-35efd76d0d72",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005287499999999999,
      "cost_per_success_usd": 0.00017624999999999997,
      "octane_score": 16.70921985815603,
      "ci_low": 16.4822695035461,
      "ci_high": 16.97021276595745,
      "ci_metric": "octane_score",
      "median_latency_ms": 778.9075002074242,
      "p95_latency_ms": 863.0846504122019,
      "input_tokens": 309,
      "output_tokens": 66,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a10a5f4d-bcbd-4973-9d33-306041118b2a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00020025,
      "cost_per_success_usd": 6.675e-05,
      "octane_score": 25.013732833957558,
      "ci_low": 24.928838951310862,
      "ci_high": 25.183520599250944,
      "ci_metric": "octane_score",
      "median_latency_ms": 723.424999974668,
      "p95_latency_ms": 743.5713132843375,
      "input_tokens": 177,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "aeae350c-f9c5-4c3e-b547-7f598aa83ef5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00040275,
      "cost_per_success_usd": 0.00013425,
      "octane_score": 17.44009931719429,
      "ci_low": 15.378215654077724,
      "ci_high": 19.747172322022617,
      "ci_metric": "octane_score",
      "median_latency_ms": 685.6772918254137,
      "p95_latency_ms": 1118.524116743356,
      "input_tokens": 375,
      "output_tokens": 27,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "526aa5c8-2aa4-449f-8ef1-fd21ac95781d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00032625000000000004,
      "cost_per_success_usd": 0.00010875,
      "octane_score": 89.94022988505746,
      "ci_low": 69.69157784743997,
      "ci_high": 117.88505747126436,
      "ci_metric": "octane_score",
      "median_latency_ms": 727.584583684802,
      "p95_latency_ms": 757.951145991683,
      "input_tokens": 159,
      "output_tokens": 46,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9e704399-670e-4243-b29b-f7cff3be0ac4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00029925000000000004,
      "cost_per_success_usd": 9.975000000000001e-05,
      "octane_score": 30.46616541353383,
      "ci_low": 27.909774436090228,
      "ci_high": 32.68170426065162,
      "ci_metric": "octane_score",
      "median_latency_ms": 565.0037499144673,
      "p95_latency_ms": 733.1737004220485,
      "input_tokens": 255,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "158b3360-f9cd-42b5-b0bf-9ee4f5631c5d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0002814,
      "cost_per_success_usd": 9.38e-05,
      "octane_score": 104.27505330490405,
      "ci_low": 80.7451403887689,
      "ci_high": 136.67377398720683,
      "ci_metric": "octane_score",
      "median_latency_ms": 2237.1675837785006,
      "p95_latency_ms": 2246.225895639509,
      "input_tokens": 159,
      "output_tokens": 51,
      "reasoning_tokens": 157,
      "undefined_reason": null
    },
    {
      "snapshot_id": "503a7a1c-a432-41d7-b637-d8cf8871cd91",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001878,
      "cost_per_success_usd": 6.26e-05,
      "octane_score": 48.546325878594246,
      "ci_low": 44.47284345047924,
      "ci_high": 52.42076502732241,
      "ci_metric": "octane_score",
      "median_latency_ms": 2169.784207828343,
      "p95_latency_ms": 2187.8904840908945,
      "input_tokens": 255,
      "output_tokens": 30,
      "reasoning_tokens": 84,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ad573c46-3bb9-48de-96dd-e91571f3ffdb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001686,
      "cost_per_success_usd": 5.6200000000000004e-05,
      "octane_score": 41.660735468564646,
      "ci_low": 37.19416386083052,
      "ci_high": 46.66666666666666,
      "ci_metric": "octane_score",
      "median_latency_ms": 1675.7381670176983,
      "p95_latency_ms": 2149.202817119658,
      "input_tokens": 375,
      "output_tokens": 27,
      "reasoning_tokens": 51,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1ee8ea2c-024b-4487-80d2-0cb970754dc0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00030480000000000004,
      "cost_per_success_usd": 0.00010160000000000002,
      "octane_score": 30.459317585301832,
      "ci_low": 24.21052631578947,
      "ci_high": 35.367063492063494,
      "ci_metric": "octane_score",
      "median_latency_ms": 1823.91441706568,
      "p95_latency_ms": 1865.2768419124186,
      "input_tokens": 288,
      "output_tokens": 114,
      "reasoning_tokens": 92,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ab3771df-21da-4a68-be86-5071c21213a6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0002898,
      "cost_per_success_usd": 9.66e-05,
      "octane_score": 30.486542443064177,
      "ci_low": 28.926940639269397,
      "ci_high": 31.837972876516776,
      "ci_metric": "octane_score",
      "median_latency_ms": 1892.2611670568585,
      "p95_latency_ms": 2011.462905164808,
      "input_tokens": 309,
      "output_tokens": 90,
      "reasoning_tokens": 100,
      "undefined_reason": null
    },
    {
      "snapshot_id": "657c72dc-eac6-420b-ba1f-7c7427d53b34",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001062,
      "cost_per_success_usd": 3.54e-05,
      "octane_score": 47.16572504708099,
      "ci_low": 46.480446927374295,
      "ci_high": 48.25626204238922,
      "ci_metric": "octane_score",
      "median_latency_ms": 1067.1783750876784,
      "p95_latency_ms": 1265.319112036377,
      "input_tokens": 177,
      "output_tokens": 21,
      "reasoning_tokens": 38,
      "undefined_reason": null
    },
    {
      "snapshot_id": "95653bce-4e26-4eb3-b29e-159200195ccb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0022164,
      "cost_per_success_usd": 0.0007388,
      "octane_score": 4.188774589424292,
      "ci_low": 3.271931424402699,
      "ci_high": 4.825392528424472,
      "ci_metric": "octane_score",
      "median_latency_ms": 3394.8849998414516,
      "p95_latency_ms": 4008.5771868005395,
      "input_tokens": 6120,
      "output_tokens": 393,
      "reasoning_tokens": 434,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6c4cca9c-ee50-4283-8819-0877b176344f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0020058,
      "cost_per_success_usd": 0.0006685999999999999,
      "octane_score": 4.404726293748131,
      "ci_low": 4.331616991781365,
      "ci_high": 4.487424547283702,
      "ci_metric": "octane_score",
      "median_latency_ms": 3325.575541704893,
      "p95_latency_ms": 3442.977840267122,
      "input_tokens": 6111,
      "output_tokens": 321,
      "reasoning_tokens": 332,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ab7ac474-f1e9-4b98-8f7e-1a449e55e2af",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0013624000000000002,
      "cost_per_success_usd": 0.0004541333333333334,
      "octane_score": 3.676600117439812,
      "ci_low": 3.5284587207664138,
      "ci_high": 3.8429561200923783,
      "ci_metric": "octane_score",
      "median_latency_ms": 2084.642625413835,
      "p95_latency_ms": 2343.172274902463,
      "input_tokens": 5216,
      "output_tokens": 112,
      "reasoning_tokens": 154,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0f1cdf6a-d6fa-47ab-a895-2a4d2516a47f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016854,
      "cost_per_success_usd": 0.0005618,
      "octane_score": 4.167556663106681,
      "ci_low": 3.8487435373331724,
      "ci_high": 4.503381986472054,
      "ci_metric": "octane_score",
      "median_latency_ms": 2413.3251244202256,
      "p95_latency_ms": 2842.8639248013496,
      "input_tokens": 6099,
      "output_tokens": 132,
      "reasoning_tokens": 256,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2bfa6acf-c448-4a3a-aac9-957cebb6c1ea",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0020392,
      "cost_per_success_usd": 0.0006797333333333334,
      "octane_score": 14.389466457434285,
      "ci_low": 11.065015785319654,
      "ci_high": 18.860337387210667,
      "ci_metric": "octane_score",
      "median_latency_ms": 3626.9984589889646,
      "p95_latency_ms": 3926.0941083543003,
      "input_tokens": 5612,
      "output_tokens": 204,
      "reasoning_tokens": 560,
      "undefined_reason": null
    },
    {
      "snapshot_id": "455c5f24-1b25-4f13-8e71-4b349d35379f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0017158,
      "cost_per_success_usd": 0.0005719333333333333,
      "octane_score": 5.313556358549948,
      "ci_low": 4.875656742556918,
      "ci_high": 5.67996678386038,
      "ci_metric": "octane_score",
      "median_latency_ms": 3507.7327080070972,
      "p95_latency_ms": 3705.3929951041937,
      "input_tokens": 5723,
      "output_tokens": 141,
      "reasoning_tokens": 335,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d5480f40-b1f1-4191-a4d3-b8d3b1f624e6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001458,
      "cost_per_success_usd": 0.00048600000000000005,
      "octane_score": 6.0596707818930025,
      "ci_low": 5.977366255144031,
      "ci_high": 6.15432098765432,
      "ci_metric": "octane_score",
      "median_latency_ms": 1250.5318336188793,
      "p95_latency_ms": 1295.0364956632257,
      "input_tokens": 309,
      "output_tokens": 84,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9a9dc6ed-6047-4dce-bc1f-e4a963744395",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00075,
      "cost_per_success_usd": 0.00025,
      "octane_score": 12.156,
      "ci_low": 11.136000000000001,
      "ci_high": 13.04,
      "ci_metric": "octane_score",
      "median_latency_ms": 1269.4122502580285,
      "p95_latency_ms": 1474.228287115693,
      "input_tokens": 255,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f5cf383e-9cf0-46cd-87fd-5b7a2de2fd14",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.000768,
      "cost_per_success_usd": 0.000256,
      "octane_score": 38.20703125,
      "ci_low": 29.207031250000004,
      "ci_high": 50.078125,
      "ci_metric": "octane_score",
      "median_latency_ms": 1129.7575421631336,
      "p95_latency_ms": 1418.0380419827995,
      "input_tokens": 159,
      "output_tokens": 45,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "391a50b6-0e22-4e9d-942f-174e5a4302e3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0009599999999999999,
      "cost_per_success_usd": 0.00031999999999999997,
      "octane_score": 7.316666666666667,
      "ci_low": 6.668750000000001,
      "ci_high": 7.906250000000001,
      "ci_metric": "octane_score",
      "median_latency_ms": 820.6662079319358,
      "p95_latency_ms": 1134.0333080850542,
      "input_tokens": 375,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8f2416e8-3a81-4242-8f9d-d1cbf4627af3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000504,
      "cost_per_success_usd": 0.000168,
      "octane_score": 9.938492063492065,
      "ci_low": 9.904761904761905,
      "ci_high": 10.005952380952383,
      "ci_metric": "octane_score",
      "median_latency_ms": 1161.851542070508,
      "p95_latency_ms": 2693.222016934305,
      "input_tokens": 177,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6a6fface-0eb7-473d-a60b-d342983345ee",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001656,
      "cost_per_success_usd": 0.000552,
      "octane_score": 5.6062801932367154,
      "ci_low": 4.333333333333334,
      "ci_high": 6.458333333333334,
      "ci_metric": "octane_score",
      "median_latency_ms": 1411.1860422417521,
      "p95_latency_ms": 1572.5507163442671,
      "input_tokens": 288,
      "output_tokens": 108,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fb4506ae-7974-4caf-9105-b433acb5251f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.027276,
      "cost_per_success_usd": 0.009092000000000001,
      "octane_score": 0.3403724886346971,
      "ci_low": 0.22940449715838893,
      "ci_high": 0.5056820929445331,
      "ci_metric": "octane_score",
      "median_latency_ms": 3291.3812920451164,
      "p95_latency_ms": 3544.914629869163,
      "input_tokens": 5880,
      "output_tokens": 369,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f45cd86b-0ddd-45ca-8b05-c96019515224",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.031839000000000006,
      "cost_per_success_usd": 0.010613000000000003,
      "octane_score": 0.27748987091303107,
      "ci_low": 0.22810585321340113,
      "ci_high": 0.4697718934439304,
      "ci_metric": "octane_score",
      "median_latency_ms": 3141.0927502438426,
      "p95_latency_ms": 3756.4391368068755,
      "input_tokens": 5889,
      "output_tokens": 297,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "161e5e82-b04d-412a-992c-e021013ab164",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.021147,
      "cost_per_success_usd": 0.007049,
      "octane_score": 1.3875727053482765,
      "ci_low": 0.8806208376905719,
      "ci_high": 2.3061733320035898,
      "ci_metric": "octane_score",
      "median_latency_ms": 3743.5884177684784,
      "p95_latency_ms": 4233.211367204785,
      "input_tokens": 5166,
      "output_tokens": 182,
      "reasoning_tokens": 69,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6e07a951-bd8b-4727-818a-3a77dce21dab",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.016449,
      "cost_per_success_usd": 0.005482999999999999,
      "octane_score": 0.3045169919144022,
      "ci_low": 0.17603233175188898,
      "ci_high": 0.4811411066436915,
      "ci_metric": "octane_score",
      "median_latency_ms": 2534.8332496359944,
      "p95_latency_ms": 2974.277900531888,
      "input_tokens": 5151,
      "output_tokens": 90,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "90f4f62d-86d9-4e93-b63a-7d352086cf9c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.018639,
      "cost_per_success_usd": 0.006213,
      "octane_score": 0.48913568324480927,
      "ci_low": 0.3005167958656331,
      "ci_high": 0.7687622711063029,
      "ci_metric": "octane_score",
      "median_latency_ms": 2688.2995003834367,
      "p95_latency_ms": 3531.8114878609776,
      "input_tokens": 5490,
      "output_tokens": 117,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d412bb3e-98d4-4f49-83d7-914b3735940b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.020415,
      "cost_per_success_usd": 0.006804999999999999,
      "octane_score": 0.3440607396522165,
      "ci_low": 0.21895786135024933,
      "ci_high": 0.5273631840796019,
      "ci_metric": "octane_score",
      "median_latency_ms": 2246.033415198326,
      "p95_latency_ms": 3907.347541674971,
      "input_tokens": 5973,
      "output_tokens": 108,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5a1490ed-d619-451e-b4f9-9ef35f4062ec",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0017100000000000001,
      "cost_per_success_usd": 0.0005700000000000001,
      "octane_score": 5.3315789473684205,
      "ci_low": 4.7204520187745205,
      "ci_high": 6.033333333333334,
      "ci_metric": "octane_score",
      "median_latency_ms": 1936.5352503955364,
      "p95_latency_ms": 2188.8402379117906,
      "input_tokens": 255,
      "output_tokens": 30,
      "reasoning_tokens": 70,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9ab4bbf5-15d6-498f-9b07-20996fc61cf9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001872,
      "cost_per_success_usd": 0.000624,
      "octane_score": 4.9594017094017095,
      "ci_low": 3.8333333333333335,
      "ci_high": 5.7131410256410255,
      "ci_metric": "octane_score",
      "median_latency_ms": 1007.7552506700158,
      "p95_latency_ms": 1094.7423878125846,
      "input_tokens": 288,
      "output_tokens": 108,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8c457785-799b-4358-b6d8-67ed4b83220d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016259999999999998,
      "cost_per_success_usd": 0.000542,
      "octane_score": 5.4335793357933575,
      "ci_low": 5.3597785977859775,
      "ci_high": 5.518450184501845,
      "ci_metric": "octane_score",
      "median_latency_ms": 919.2689163610339,
      "p95_latency_ms": 1563.7687792070208,
      "input_tokens": 309,
      "output_tokens": 84,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "919071b3-5fb7-426a-b464-4b9699ef6a91",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000534,
      "cost_per_success_usd": 0.000178,
      "octane_score": 9.380149812734084,
      "ci_low": 9.348314606741573,
      "ci_high": 9.443820224719104,
      "ci_metric": "octane_score",
      "median_latency_ms": 858.0120829865336,
      "p95_latency_ms": 966.5267336182296,
      "input_tokens": 177,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4e5fba07-6335-478e-9740-73495b56593f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0010019999999999999,
      "cost_per_success_usd": 0.00033399999999999993,
      "octane_score": 7.009980039920161,
      "ci_low": 6.389221556886229,
      "ci_high": 7.5748502994011995,
      "ci_metric": "octane_score",
      "median_latency_ms": 879.3912921100855,
      "p95_latency_ms": 962.0162789709866,
      "input_tokens": 375,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8176e2b7-a581-43c5-855a-b29e2c6aac2b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00231,
      "cost_per_success_usd": 0.0007700000000000001,
      "octane_score": 12.702597402597402,
      "ci_low": 9.81233595800525,
      "ci_high": 16.649350649350648,
      "ci_metric": "octane_score",
      "median_latency_ms": 1987.3500000685453,
      "p95_latency_ms": 2430.09536229074,
      "input_tokens": 159,
      "output_tokens": 53,
      "reasoning_tokens": 113,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ea2eb83b-e2d6-480f-b7c9-9fc0871752d8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0062412,
      "cost_per_success_usd": 0.0020804,
      "octane_score": 1.487534448503493,
      "ci_low": 1.0775301764159704,
      "ci_high": 2.0593564141951237,
      "ci_metric": "octane_score",
      "median_latency_ms": 2409.236707724631,
      "p95_latency_ms": 2607.0347698405385,
      "input_tokens": 5880,
      "output_tokens": 375,
      "reasoning_tokens": 125,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f877c01d-6db2-423e-ad2e-53ef669f6dcf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.008202,
      "cost_per_success_usd": 0.0027339999999999995,
      "octane_score": 1.0771762984637894,
      "ci_low": 1.0583041006411031,
      "ci_high": 1.0974289580514207,
      "ci_metric": "octane_score",
      "median_latency_ms": 2192.575625143945,
      "p95_latency_ms": 3040.982562769204,
      "input_tokens": 5904,
      "output_tokens": 318,
      "reasoning_tokens": 65,
      "undefined_reason": null
    },
    {
      "snapshot_id": "851654b3-9520-468e-9ce2-1b0566fb77ea",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0038058,
      "cost_per_success_usd": 0.0012686,
      "octane_score": 1.3161490356823797,
      "ci_low": 0.7947009360621927,
      "ci_high": 2.15607782369146,
      "ci_metric": "octane_score",
      "median_latency_ms": 2315.9980829805136,
      "p95_latency_ms": 2376.5490705147386,
      "input_tokens": 5151,
      "output_tokens": 96,
      "reasoning_tokens": 80,
      "undefined_reason": null
    },
    {
      "snapshot_id": "60ba759a-339f-4407-8999-7b8cbc582280",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004449,
      "cost_per_success_usd": 0.001483,
      "octane_score": 1.5787817487075748,
      "ci_low": 1.046427024686012,
      "ci_high": 2.3289391086001254,
      "ci_metric": "octane_score",
      "median_latency_ms": 1849.8386247083545,
      "p95_latency_ms": 2159.342136885971,
      "input_tokens": 5973,
      "output_tokens": 114,
      "reasoning_tokens": 55,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f2824abe-210f-4a00-b744-f1d32878f5ed",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.009371,
      "cost_per_success_usd": 0.003123666666666667,
      "octane_score": 3.1312560025610923,
      "ci_low": 2.415311726068698,
      "ci_high": 4.067689053410893,
      "ci_metric": "octane_score",
      "median_latency_ms": 3383.1026665866375,
      "p95_latency_ms": 3623.0361410416663,
      "input_tokens": 5471,
      "output_tokens": 210,
      "reasoning_tokens": 440,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fe625a31-0573-4ae4-90f6-00ef6bb7c6d6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.008187,
      "cost_per_success_usd": 0.002729,
      "octane_score": 1.1135947233418835,
      "ci_low": 1.0246595509753404,
      "ci_high": 1.190937306021105,
      "ci_metric": "octane_score",
      "median_latency_ms": 3106.6724164411426,
      "p95_latency_ms": 3364.2741666175425,
      "input_tokens": 5661,
      "output_tokens": 141,
      "reasoning_tokens": 280,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3cbbf5f6-ebe9-479a-87fc-e146df1d097c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 8.834999999999999e-05,
      "cost_per_success_usd": 2.9449999999999998e-05,
      "octane_score": 100.0,
      "ci_low": 2.9049999999999998e-05,
      "ci_high": 2.991e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 1060.7846667990088,
      "p95_latency_ms": 2517.519528418779,
      "input_tokens": 514,
      "output_tokens": 327,
      "reasoning_tokens": 102,
      "undefined_reason": null
    },
    {
      "snapshot_id": "729ffcb3-e4a5-42ce-add5-12a6604e266a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 7.023999999999999e-05,
      "cost_per_success_usd": 2.341333333333333e-05,
      "octane_score": 100.0,
      "ci_low": 2.134e-05,
      "ci_high": 2.5300000000000002e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 1662.4722499400377,
      "p95_latency_ms": 2202.7145377360284,
      "input_tokens": 545,
      "output_tokens": 22,
      "reasoning_tokens": 295,
      "undefined_reason": null
    },
    {
      "snapshot_id": "55c00991-72c6-4b9d-bf43-f146d7dd9aa4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 5.009e-05,
      "cost_per_success_usd": 1.669666666666667e-05,
      "octane_score": 100.0,
      "ci_low": 1.664e-05,
      "ci_high": 1.6810000000000003e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 1786.806708201766,
      "p95_latency_ms": 1861.8029955774546,
      "input_tokens": 321,
      "output_tokens": 15,
      "reasoning_tokens": 223,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1313ec7d-60d9-4c44-ae73-44f31a681e7e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00029343,
      "cost_per_success_usd": 9.781e-05,
      "octane_score": 100.0,
      "ci_low": 7.477e-05,
      "ci_high": 0.0001282,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 11933.56404080987,
      "p95_latency_ms": 15405.860453657806,
      "input_tokens": 329,
      "output_tokens": 733,
      "reasoning_tokens": 935,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b5922637-260c-49b5-86f3-0f5452a8a6dc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 9.117000000000001e-05,
      "cost_per_success_usd": 3.0390000000000002e-05,
      "octane_score": 100.0,
      "ci_low": 2.7840000000000005e-05,
      "ci_high": 3.26e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 4878.946959041059,
      "p95_latency_ms": 5462.013346515596,
      "input_tokens": 438,
      "output_tokens": 198,
      "reasoning_tokens": 261,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e444ea96-8a2f-41f1-83e3-ff1ce96bcde1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 9.284e-05,
      "cost_per_success_usd": 3.0946666666666666e-05,
      "octane_score": 100.0,
      "ci_low": 2.392e-05,
      "ci_high": 3.565e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 5353.179832920432,
      "p95_latency_ms": 9914.371095597742,
      "input_tokens": 471,
      "output_tokens": 169,
      "reasoning_tokens": 294,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fb204c15-d811-4532-a06d-caf136389e47",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0041396,
      "cost_per_success_usd": 0.0013798666666666667,
      "octane_score": 2.1342641801140205,
      "ci_low": 1.8124898025779081,
      "ci_high": 2.5511088011088003,
      "ci_metric": "octane_score",
      "median_latency_ms": 2883.0665424466133,
      "p95_latency_ms": 3416.386816557497,
      "input_tokens": 915,
      "output_tokens": 90,
      "reasoning_tokens": 440,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bd5b140b-a58f-4e7c-9a83-6b4642614141",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003952,
      "cost_per_success_usd": 0.0013173333333333333,
      "octane_score": 2.3491902834008096,
      "ci_low": 1.8157894736842106,
      "ci_high": 2.9370295472418833,
      "ci_metric": "octane_score",
      "median_latency_ms": 2793.8942909240723,
      "p95_latency_ms": 3342.1951030381024,
      "input_tokens": 888,
      "output_tokens": 108,
      "reasoning_tokens": 436,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a707aaa5-bd04-465f-849c-cfb6ada3fac9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004612,
      "cost_per_success_usd": 0.0015373333333333333,
      "octane_score": 1.9767996530789247,
      "ci_low": 1.7847405816259092,
      "ci_high": 2.1889832055494707,
      "ci_metric": "octane_score",
      "median_latency_ms": 3556.0121247544885,
      "p95_latency_ms": 3683.3067001774907,
      "input_tokens": 858,
      "output_tokens": 12,
      "reasoning_tokens": 652,
      "undefined_reason": null
    },
    {
      "snapshot_id": "040e3411-aba9-43b0-8230-3d4e5c2480f3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0065112,
      "cost_per_success_usd": 0.0021704,
      "octane_score": 4.506542572797641,
      "ci_low": 3.4449870991522302,
      "ci_high": 5.85416059744804,
      "ci_metric": "octane_score",
      "median_latency_ms": 3565.0310833007097,
      "p95_latency_ms": 3577.1194338798523,
      "input_tokens": 768,
      "output_tokens": 42,
      "reasoning_tokens": 896,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b5686ff4-7d61-465a-a1fb-498318b9bed6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0036836000000000004,
      "cost_per_success_usd": 0.0012278666666666667,
      "octane_score": 1.9068302747312411,
      "ci_low": 1.7147883680016556,
      "ci_high": 2.1445086705202305,
      "ci_metric": "octane_score",
      "median_latency_ms": 2707.137208431959,
      "p95_latency_ms": 2707.9480703920126,
      "input_tokens": 969,
      "output_tokens": 6,
      "reasoning_tokens": 430,
      "undefined_reason": null
    },
    {
      "snapshot_id": "acc1c172-f612-4d4a-aca5-bc4c22f41fb0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003634,
      "cost_per_success_usd": 0.0012113333333333334,
      "octane_score": 1.3783709411117229,
      "ci_low": 1.2959225913277455,
      "ci_high": 1.54198990272134,
      "ci_metric": "octane_score",
      "median_latency_ms": 2435.2670842781663,
      "p95_latency_ms": 2783.5098217241466,
      "input_tokens": 783,
      "output_tokens": 3,
      "reasoning_tokens": 523,
      "undefined_reason": null
    },
    {
      "snapshot_id": "99c873c8-303c-4a50-a92f-061bd271ed02",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0064860000000000004,
      "cost_per_success_usd": 0.0021620000000000003,
      "octane_score": 1.0829478877582484,
      "ci_low": 0.9810538780343397,
      "ci_high": 1.1936936936936933,
      "ci_metric": "octane_score",
      "median_latency_ms": 4228.3580005168915,
      "p95_latency_ms": 4811.114151123911,
      "input_tokens": 969,
      "output_tokens": 6,
      "reasoning_tokens": 912,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5c8c3e69-47fc-4b64-9e50-298a9db90d98",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.005748,
      "cost_per_success_usd": 0.001916,
      "octane_score": 1.586116910229645,
      "ci_low": 1.4530271398747394,
      "ci_high": 1.7284684684684686,
      "ci_metric": "octane_score",
      "median_latency_ms": 2816.1112498492002,
      "p95_latency_ms": 3169.207349792123,
      "input_tokens": 858,
      "output_tokens": 12,
      "reasoning_tokens": 788,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8ab54256-2880-4b2b-83f0-0f0a755fab63",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.007566000000000001,
      "cost_per_success_usd": 0.0025220000000000004,
      "octane_score": 1.1677240285487704,
      "ci_low": 1.093318540433925,
      "ci_high": 1.2257488320967298,
      "ci_metric": "octane_score",
      "median_latency_ms": 3832.6124167069793,
      "p95_latency_ms": 4577.925291378051,
      "input_tokens": 915,
      "output_tokens": 90,
      "reasoning_tokens": 1026,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0de4e410-a77d-4d46-a7da-ae9cbce4b999",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004194,
      "cost_per_success_usd": 0.001398,
      "octane_score": 2.2136385312350977,
      "ci_low": 1.7797619047619042,
      "ci_high": 2.5935019841269837,
      "ci_metric": "octane_score",
      "median_latency_ms": 2771.5150425210595,
      "p95_latency_ms": 2940.0982786901295,
      "input_tokens": 888,
      "output_tokens": 108,
      "reasoning_tokens": 455,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2a075ef0-8abf-41aa-b3bb-a06b1abd1446",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.008591999999999999,
      "cost_per_success_usd": 0.002864,
      "octane_score": 3.4151536312849164,
      "ci_low": 2.65137146206855,
      "ci_high": 4.476256983240224,
      "ci_metric": "octane_score",
      "median_latency_ms": 6339.018958620727,
      "p95_latency_ms": 7360.872733406723,
      "input_tokens": 768,
      "output_tokens": 42,
      "reasoning_tokens": 1230,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ba1d564f-d716-4bba-8def-7a0c2128457f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00705,
      "cost_per_success_usd": 0.00235,
      "octane_score": 0.7104964539007094,
      "ci_low": 0.7068819031435855,
      "ci_high": 0.7153191489361703,
      "ci_metric": "octane_score",
      "median_latency_ms": 6433.145250193775,
      "p95_latency_ms": 18288.5015623644,
      "input_tokens": 783,
      "output_tokens": 3,
      "reasoning_tokens": 1039,
      "undefined_reason": null
    },
    {
      "snapshot_id": "661f5c75-9bf3-403c-b810-a8762ca8da00",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00292275,
      "cost_per_success_usd": 0.00097425,
      "octane_score": 3.1193225558121633,
      "ci_low": 1.9435566418081929,
      "ci_high": 4.871521430022344,
      "ci_metric": "octane_score",
      "median_latency_ms": 5149.050292558968,
      "p95_latency_ms": 10539.33692993596,
      "input_tokens": 260,
      "output_tokens": 79,
      "reasoning_tokens": 1168,
      "undefined_reason": null
    },
    {
      "snapshot_id": "971cb343-e942-4925-9215-e284380e0b7b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00376739,
      "cost_per_success_usd": 0.0012557966666666667,
      "octane_score": 7.788681288637491,
      "ci_low": 5.912145517608341,
      "ci_high": 10.208659045121422,
      "ci_metric": "octane_score",
      "median_latency_ms": 10262.573125772178,
      "p95_latency_ms": 11481.523937173188,
      "input_tokens": 164,
      "output_tokens": 399,
      "reasoning_tokens": 1248,
      "undefined_reason": null
    },
    {
      "snapshot_id": "92c14e6a-8303-4bc8-b019-c1b8a0267296",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00384417,
      "cost_per_success_usd": 0.00128139,
      "octane_score": 1.8271824607132356,
      "ci_low": 1.2981947130883307,
      "ci_high": 2.50644345132534,
      "ci_metric": "octane_score",
      "median_latency_ms": 10985.17158254981,
      "p95_latency_ms": 15348.467320855707,
      "input_tokens": 379,
      "output_tokens": 13,
      "reasoning_tokens": 1641,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5e51d9a7-9363-401f-8d89-20504a59b601",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0020254500000000003,
      "cost_per_success_usd": 0.0006751500000000001,
      "octane_score": 2.4730306845392382,
      "ci_low": 1.5046108557868494,
      "ci_high": 41.08641975308642,
      "ci_metric": "octane_score",
      "median_latency_ms": 9234.773458912969,
      "p95_latency_ms": 44047.036107257,
      "input_tokens": 201,
      "output_tokens": 5,
      "reasoning_tokens": 855,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fbc751ac-7d93-4d3a-b5ee-b595baac8727",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0038236299999999997,
      "cost_per_success_usd": 0.0012745433333333333,
      "octane_score": 2.4280592002887307,
      "ci_low": 1.6188881913928994,
      "ci_high": 3.8486168975517834,
      "ci_metric": "octane_score",
      "median_latency_ms": 8570.128666236997,
      "p95_latency_ms": 13060.337866563348,
      "input_tokens": 296,
      "output_tokens": 119,
      "reasoning_tokens": 1532,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4252bbbd-563e-475b-b8df-a68bf08fa56a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00503108,
      "cost_per_success_usd": 0.0016770266666666668,
      "octane_score": 1.756084180732566,
      "ci_low": 1.3065371220691644,
      "ci_high": 3.1400293987526933,
      "ci_metric": "octane_score",
      "median_latency_ms": 16321.229167282581,
      "p95_latency_ms": 16664.396166149527,
      "input_tokens": 313,
      "output_tokens": 85,
      "reasoning_tokens": 2106,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5f610f43-ad64-4cb1-b281-6ea996198db1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00813,
      "cost_per_success_usd": 0.00271,
      "octane_score": 1.1419434194341942,
      "ci_low": 0.8867798620894193,
      "ci_high": 1.3154981549815496,
      "ci_metric": "octane_score",
      "median_latency_ms": 2147.4204575642943,
      "p95_latency_ms": 2252.6989333331585,
      "input_tokens": 528,
      "output_tokens": 200,
      "reasoning_tokens": 294,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d9c2690b-7a8f-4807-b4f7-ac8b80cfeb01",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.008004,
      "cost_per_success_usd": 0.0026680000000000002,
      "octane_score": 1.1038230884557718,
      "ci_low": 1.0839775993656455,
      "ci_high": 1.124004636629372,
      "ci_metric": "octane_score",
      "median_latency_ms": 1341.9351251795888,
      "p95_latency_ms": 1397.540349792689,
      "input_tokens": 546,
      "output_tokens": 218,
      "reasoning_tokens": 264,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f9c6b236-fbcb-449d-ba85-1b69026285eb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.020163,
      "cost_per_success_usd": 0.006721,
      "octane_score": 0.24842533353171653,
      "ci_low": 0.2182228495747944,
      "ci_high": 0.2685042240233287,
      "ci_metric": "octane_score",
      "median_latency_ms": 5273.353541269898,
      "p95_latency_ms": 5817.070766631514,
      "input_tokens": 414,
      "output_tokens": 0,
      "reasoning_tokens": 1319,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ec9c8d1a-9d3b-476a-9085-a152f8aa9703",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007346999999999999,
      "cost_per_success_usd": 0.002449,
      "octane_score": 0.9560364774737989,
      "ci_low": 0.8022270636649722,
      "ci_high": 1.180037313432836,
      "ci_metric": "octane_score",
      "median_latency_ms": 2539.496541954577,
      "p95_latency_ms": 2582.5385157950222,
      "input_tokens": 612,
      "output_tokens": 8,
      "reasoning_tokens": 417,
      "undefined_reason": null
    },
    {
      "snapshot_id": "96fdecda-a002-416f-b535-0f0a2d1f395f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.043343999999999994,
      "cost_per_success_usd": 0.014447999999999997,
      "octane_score": 0.6769795127353267,
      "ci_low": 0.5213220727765614,
      "ci_high": 0.8873200442967887,
      "ci_metric": "octane_score",
      "median_latency_ms": 4584.47637502104,
      "p95_latency_ms": 4817.253200337291,
      "input_tokens": 396,
      "output_tokens": 1377,
      "reasoning_tokens": 1491,
      "undefined_reason": null
    },
    {
      "snapshot_id": "806cc8b3-acb7-4ec3-b04e-e31ac2e9163a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007391999999999999,
      "cost_per_success_usd": 0.0024639999999999996,
      "octane_score": 1.2333603896103897,
      "ci_low": 1.1266265091658194,
      "ci_high": 1.3484299007618568,
      "ci_metric": "octane_score",
      "median_latency_ms": 2223.6690409481525,
      "p95_latency_ms": 2402.1719292737544,
      "input_tokens": 492,
      "output_tokens": 170,
      "reasoning_tokens": 282,
      "undefined_reason": null
    },
    {
      "snapshot_id": "56c8adff-684d-4a76-8550-f908a6ab3625",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0006988000000000001,
      "cost_per_success_usd": 0.00023293333333333332,
      "octane_score": 12.643102461362334,
      "ci_low": 12.011257550796266,
      "ci_high": 13.760162601626012,
      "ci_metric": "octane_score",
      "median_latency_ms": 12028.308416716754,
      "p95_latency_ms": 14913.51210437715,
      "input_tokens": 318,
      "output_tokens": 794,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e00410c7-b8d5-495a-b482-9b5d5d97af7f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00015539999999999998,
      "cost_per_success_usd": 5.179999999999999e-05,
      "octane_score": 59.74259974259975,
      "ci_low": 46.17760617760619,
      "ci_high": 68.82239382239383,
      "ci_metric": "octane_score",
      "median_latency_ms": 677.3408325389028,
      "p95_latency_ms": 2134.5495080575347,
      "input_tokens": 297,
      "output_tokens": 120,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9e95dfe3-06d6-4b33-9de4-2fd2b3b34ff0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.0007334,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 9721.74808382988,
      "p95_latency_ms": 9862.958683818579,
      "input_tokens": 267,
      "output_tokens": 850,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "9d9dbdee-0c52-4ab4-b3fd-ab44d5908ecb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0007646,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 10426.35695822537,
      "p95_latency_ms": 12151.460008136928,
      "input_tokens": 171,
      "output_tokens": 913,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "613cdcd1-6791-413d-9a5b-25433dca1b9d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.0003624,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 1382.0071248337626,
      "p95_latency_ms": 1593.304550461471,
      "input_tokens": 384,
      "output_tokens": 357,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "40d6a438-af3c-431a-9bc0-01c2f1e1cd6c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 4.2e-05,
      "cost_per_success_usd": 1.4e-05,
      "octane_score": 119.26190476190479,
      "ci_low": 118.85714285714285,
      "ci_high": 120.0714285714286,
      "ci_metric": "octane_score",
      "median_latency_ms": 359.1777915135026,
      "p95_latency_ms": 627.2140287794172,
      "input_tokens": 186,
      "output_tokens": 6,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "91f2d760-df56-489d-9672-f73061c72ae1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 24.434749968349934,
      "p95_latency_ms": 28.6828619427979,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "abd20a8b-1ba1-4931-93de-f105bd314e3a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 28.239957988262177,
      "p95_latency_ms": 30.510882940143347,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "31789e88-c94b-479f-8d8d-6a0d75d3a076",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 33.794707618653774,
      "p95_latency_ms": 45.26940817013382,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "602998d2-8274-4913-bc95-ed332540633d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.645208433270454,
      "p95_latency_ms": 84.31700794026254,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "a2895137-3ec1-4c46-9558-0f2b19031e68",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.54716717451811,
      "p95_latency_ms": 31.068491376936436,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "3e35098a-d8e4-40f8-9624-cd0e374a0328",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 25.249292142689228,
      "p95_latency_ms": 30.368490889668465,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "930ed5c1-53c5-4f5c-92dc-0a9cfc878b59",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.01234,
      "cost_per_success_usd": 0.004113333333333333,
      "octane_score": 0.5692058346839546,
      "ci_low": 0.5209113100081367,
      "ci_high": 0.6125907990314771,
      "ci_metric": "octane_score",
      "median_latency_ms": 2497.6781252771616,
      "p95_latency_ms": 2845.238775294274,
      "input_tokens": 609,
      "output_tokens": 21,
      "reasoning_tokens": 104,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b756b0ad-4df6-44ea-a4d0-077a7cf31854",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.03087,
      "cost_per_success_usd": 0.01029,
      "octane_score": 0.30074505992873335,
      "ci_low": 0.23667546174142476,
      "ci_high": 0.3698974177573399,
      "ci_metric": "octane_score",
      "median_latency_ms": 2816.956415772438,
      "p95_latency_ms": 3026.813241187483,
      "input_tokens": 477,
      "output_tokens": 169,
      "reasoning_tokens": 353,
      "undefined_reason": null
    },
    {
      "snapshot_id": "aff73fff-f98f-4590-a057-fdefd25c0f92",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 3,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.01499,
      "cost_per_success_usd": 0.007495,
      "octane_score": 0.40547031354236157,
      "ci_low": 0.1513717421124829,
      "ci_high": 0.9360587002096434,
      "ci_metric": "octane_score",
      "median_latency_ms": 2385.6109166517854,
      "p95_latency_ms": 3010.1290917955334,
      "input_tokens": 369,
      "output_tokens": 82,
      "reasoning_tokens": 144,
      "undefined_reason": null
    },
    {
      "snapshot_id": "98f6f2a3-0035-4584-a375-1263bef98cc6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.02478,
      "cost_per_success_usd": 0.00826,
      "octane_score": 1.1841404358353511,
      "ci_low": 0.7011393514461001,
      "ci_high": 2.601329787234042,
      "ci_metric": "octane_score",
      "median_latency_ms": 3591.315290890634,
      "p95_latency_ms": 4819.201190955937,
      "input_tokens": 273,
      "output_tokens": 63,
      "reasoning_tokens": 378,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8d253369-9a57-4325-9036-33c7d9a90dcb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.011389999999999999,
      "cost_per_success_usd": 0.0037966666666666665,
      "octane_score": 0.7756804214223002,
      "ci_low": 0.6226215644820295,
      "ci_high": 0.8929929929929931,
      "ci_metric": "octane_score",
      "median_latency_ms": 1443.2877078652382,
      "p95_latency_ms": 2519.481457304209,
      "input_tokens": 459,
      "output_tokens": 108,
      "reasoning_tokens": 28,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4e2bfc53-c485-4ee3-944e-c0b235b9ef05",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 3,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.013510000000000001,
      "cost_per_success_usd": 0.006755000000000001,
      "octane_score": 0.24717493214902544,
      "ci_low": 0.07408726625111309,
      "ci_high": 1.1021929824561407,
      "ci_metric": "octane_score",
      "median_latency_ms": 2163.3686665445566,
      "p95_latency_ms": 2346.1771172471344,
      "input_tokens": 291,
      "output_tokens": 188,
      "reasoning_tokens": 24,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b2e6d4e8-b87b-441e-a2a8-1fa8dbb1c162",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000637845,
      "cost_per_success_usd": 0.00021261499999999998,
      "octane_score": 14.293441196528939,
      "ci_low": 13.09409025703737,
      "ci_high": 15.546678119099905,
      "ci_metric": "octane_score",
      "median_latency_ms": 3786.8127087131143,
      "p95_latency_ms": 4033.2642461173236,
      "input_tokens": 297,
      "output_tokens": 2379,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bafbe280-04d2-4148-ba25-f06321ef1ba8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.001127165,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 5246.110708452761,
      "p95_latency_ms": 9858.8210824877,
      "input_tokens": 369,
      "output_tokens": 4243,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "6eb2ae1c-fabb-4ead-9ceb-0ade247305be",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00058682,
      "cost_per_success_usd": 0.0001956066666666667,
      "octane_score": 11.969598854844754,
      "ci_low": 8.162014976174268,
      "ci_high": 16.936772426386664,
      "ci_metric": "octane_score",
      "median_latency_ms": 3514.4787495955825,
      "p95_latency_ms": 6994.293799996375,
      "input_tokens": 420,
      "output_tokens": 2152,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7b9dab33-ebf9-46a7-a3fb-d42da9932d7c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000481715,
      "cost_per_success_usd": 0.00016057166666666667,
      "octane_score": 10.3982645340087,
      "ci_low": 9.468182634419275,
      "ci_high": 10.972481470565764,
      "ci_metric": "octane_score",
      "median_latency_ms": 4522.312832996249,
      "p95_latency_ms": 4585.161182843149,
      "input_tokens": 195,
      "output_tokens": 1804,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c78d6547-5833-4d7c-87e6-89eafc7a8f86",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0006816550000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3891.5418330579996,
      "p95_latency_ms": 5020.941783208399,
      "input_tokens": 183,
      "output_tokens": 2576,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "210ad1f9-b758-4c83-bcc7-e132a93db1da",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0017893849999999999,
      "cost_per_success_usd": 0.0005964616666666666,
      "octane_score": 5.188374776808792,
      "ci_low": 3.937292631957069,
      "ci_high": 6.885380829314064,
      "ci_metric": "octane_score",
      "median_latency_ms": 11859.26387552172,
      "p95_latency_ms": 14122.302400320768,
      "input_tokens": 333,
      "output_tokens": 6799,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0805509d-60c4-427a-96ef-75d5030663db",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0026160000000000003,
      "cost_per_success_usd": 0.000872,
      "octane_score": 1.9147553516819573,
      "ci_low": 1.7320193637621026,
      "ci_high": 2.113502109704642,
      "ci_metric": "octane_score",
      "median_latency_ms": 2114.5458752289414,
      "p95_latency_ms": 2900.4126370884474,
      "input_tokens": 330,
      "output_tokens": 3,
      "reasoning_tokens": 323,
      "undefined_reason": null
    },
    {
      "snapshot_id": "05bded23-d2d0-4ce6-a1cb-c4d581abb31e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003522,
      "cost_per_success_usd": 0.001174,
      "octane_score": 2.6360022714366838,
      "ci_low": 2.054982817869416,
      "ci_high": 3.0415939499709133,
      "ci_metric": "octane_score",
      "median_latency_ms": 4230.782125145197,
      "p95_latency_ms": 4388.293973635882,
      "input_tokens": 477,
      "output_tokens": 146,
      "reasoning_tokens": 282,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d2d08b2e-8303-4dc0-8aca-a08490d5bce2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003456,
      "cost_per_success_usd": 0.0011519999999999998,
      "octane_score": 2.556423611111111,
      "ci_low": 2.2441919191919193,
      "ci_high": 2.9747474747474745,
      "ci_metric": "octane_score",
      "median_latency_ms": 1325.98050031811,
      "p95_latency_ms": 3733.324201125651,
      "input_tokens": 504,
      "output_tokens": 169,
      "reasoning_tokens": 239,
      "undefined_reason": null
    },
    {
      "snapshot_id": "35d16e54-d441-4138-b018-e3bdbbbfa408",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0091368,
      "cost_per_success_usd": 0.0030456000000000003,
      "octane_score": 3.2115182558444966,
      "ci_low": 2.1390374331550803,
      "ci_high": 4.917488941816944,
      "ci_metric": "octane_score",
      "median_latency_ms": 3119.4095415994525,
      "p95_latency_ms": 7383.478328678756,
      "input_tokens": 318,
      "output_tokens": 690,
      "reasoning_tokens": 746,
      "undefined_reason": null
    },
    {
      "snapshot_id": "40556ccc-f45b-45f7-9bde-af4b1ede42f3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003648,
      "cost_per_success_usd": 0.001216,
      "octane_score": 2.4991776315789473,
      "ci_low": 2.1721970839617897,
      "ci_high": 2.9277456647398847,
      "ci_metric": "octane_score",
      "median_latency_ms": 1845.6411669030786,
      "p95_latency_ms": 2535.7443296350534,
      "input_tokens": 423,
      "output_tokens": 155,
      "reasoning_tokens": 312,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fef99042-48b7-41d1-a231-f7bedc9f868b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004098,
      "cost_per_success_usd": 0.001366,
      "octane_score": 1.7140068326012687,
      "ci_low": 1.5691287878787876,
      "ci_high": 1.8440233236151604,
      "ci_metric": "octane_score",
      "median_latency_ms": 4154.975292272866,
      "p95_latency_ms": 6375.65674195066,
      "input_tokens": 555,
      "output_tokens": 14,
      "reasoning_tokens": 484,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7a93b0ba-ddc2-4b9c-b0c4-a2ccbe6c3370",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0017496999999999999,
      "cost_per_success_usd": 0.0005832333333333333,
      "octane_score": 4.0144024689946844,
      "ci_low": 3.558798196281096,
      "ci_high": 4.530744336569579,
      "ci_metric": "octane_score",
      "median_latency_ms": 2926.8439998850226,
      "p95_latency_ms": 3094.9745755642653,
      "input_tokens": 546,
      "output_tokens": 0,
      "reasoning_tokens": 470,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5f3b7777-c23b-430a-ab13-0649e7a9bcd8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0022447500000000002,
      "cost_per_success_usd": 0.0007482500000000001,
      "octane_score": 4.061476779151352,
      "ci_low": 3.7206815903775476,
      "ci_high": 4.356832609421984,
      "ci_metric": "octane_score",
      "median_latency_ms": 3307.2583340108395,
      "p95_latency_ms": 5243.170446343719,
      "input_tokens": 423,
      "output_tokens": 292,
      "reasoning_tokens": 350,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1e91f020-88cd-49dd-b8c0-fa091416f208",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0017665500000000002,
      "cost_per_success_usd": 0.00058885,
      "octane_score": 5.001273669015878,
      "ci_low": 3.9978280051584876,
      "ci_high": 6.472513329656183,
      "ci_metric": "octane_score",
      "median_latency_ms": 2706.4441246911883,
      "p95_latency_ms": 2732.8119500540197,
      "input_tokens": 491,
      "output_tokens": 203,
      "reasoning_tokens": 280,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e919d57e-7d27-4c42-b5eb-fa006b5b2ebe",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0008196500000000001,
      "cost_per_success_usd": 0.0002732166666666667,
      "octane_score": 6.111145000915025,
      "ci_low": 5.973792855860706,
      "ci_high": 6.359423601853616,
      "ci_metric": "octane_score",
      "median_latency_ms": 1962.0459591969848,
      "p95_latency_ms": 15487.564258556811,
      "input_tokens": 321,
      "output_tokens": 67,
      "reasoning_tokens": 144,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2cdf2d20-5037-48bd-9012-73b89026e238",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00373265,
      "cost_per_success_usd": 0.0012442166666666666,
      "octane_score": 7.861171017909529,
      "ci_low": 6.009403506891887,
      "ci_high": 10.303671654186704,
      "ci_metric": "octane_score",
      "median_latency_ms": 5534.077124670148,
      "p95_latency_ms": 10107.60468803346,
      "input_tokens": 309,
      "output_tokens": 661,
      "reasoning_tokens": 462,
      "undefined_reason": null
    },
    {
      "snapshot_id": "59cceca6-6abb-4bdf-a2ef-8ef54766c835",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016913499999999999,
      "cost_per_success_usd": 0.0005637833333333333,
      "octane_score": 5.489106335176043,
      "ci_low": 4.242764655452746,
      "ci_high": 6.311410108878463,
      "ci_metric": "octane_score",
      "median_latency_ms": 1895.3537922352552,
      "p95_latency_ms": 4080.0247787497933,
      "input_tokens": 459,
      "output_tokens": 223,
      "reasoning_tokens": 241,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c8d47c4d-f727-40bb-b169-a21edf655923",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.011394,
      "cost_per_success_usd": 0.0037979999999999997,
      "octane_score": 2.5753027909426014,
      "ci_low": 1.756697408871322,
      "ci_high": 4.15858843537415,
      "ci_metric": "octane_score",
      "median_latency_ms": 11960.118542425334,
      "p95_latency_ms": 12152.69255489111,
      "input_tokens": 297,
      "output_tokens": 57,
      "reasoning_tokens": 1743,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a7fef2b3-eb09-4a38-ab8b-322fe9caa212",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0034140000000000004,
      "cost_per_success_usd": 0.0011380000000000001,
      "octane_score": 2.719390743995313,
      "ci_low": 2.162748643761302,
      "ci_high": 3.154867256637168,
      "ci_metric": "octane_score",
      "median_latency_ms": 3664.3102494999766,
      "p95_latency_ms": 3981.02902546525,
      "input_tokens": 447,
      "output_tokens": 150,
      "reasoning_tokens": 270,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a6be6581-6e08-426b-89ee-0b800e86db45",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0031800000000000005,
      "cost_per_success_usd": 0.0010600000000000002,
      "octane_score": 2.7783018867924523,
      "ci_low": 2.700308641975308,
      "ci_high": 2.859294871794871,
      "ci_metric": "octane_score",
      "median_latency_ms": 2881.819000467658,
      "p95_latency_ms": 3222.151712421328,
      "input_tokens": 483,
      "output_tokens": 126,
      "reasoning_tokens": 243,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c5190b68-8e59-47ad-b16b-17d6bbed22e2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0018960000000000001,
      "cost_per_success_usd": 0.0006320000000000001,
      "octane_score": 2.641877637130802,
      "ci_low": 2.522155085599195,
      "ci_high": 2.719696969696971,
      "ci_metric": "octane_score",
      "median_latency_ms": 2005.9705832973123,
      "p95_latency_ms": 3541.3170340470965,
      "input_tokens": 309,
      "output_tokens": 18,
      "reasoning_tokens": 195,
      "undefined_reason": null
    },
    {
      "snapshot_id": "96652c49-e44a-4bbc-a536-25ac3e96a63a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0038399999999999997,
      "cost_per_success_usd": 0.0012799999999999999,
      "octane_score": 1.8291666666666668,
      "ci_low": 1.6788247213779126,
      "ci_high": 1.9704049844236762,
      "ci_metric": "octane_score",
      "median_latency_ms": 3434.6305830404162,
      "p95_latency_ms": 4051.0619960725307,
      "input_tokens": 534,
      "output_tokens": 24,
      "reasoning_tokens": 438,
      "undefined_reason": null
    },
    {
      "snapshot_id": "28c1452e-c018-4351-bd8b-ed2dcdd87834",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0033,
      "cost_per_success_usd": 0.0011,
      "octane_score": 2.7627272727272727,
      "ci_low": 2.4916378316032297,
      "ci_high": 3.1137295081967213,
      "ci_metric": "octane_score",
      "median_latency_ms": 4213.31129129976,
      "p95_latency_ms": 4549.73987871781,
      "input_tokens": 411,
      "output_tokens": 36,
      "reasoning_tokens": 377,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6d7e111d-4dfb-40e0-a473-2d21e6d33e17",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.042126,
      "cost_per_success_usd": 0.014041999999999999,
      "octane_score": 0.21642216208517306,
      "ci_low": 0.11425241550434229,
      "ci_high": 1.6864966949952784,
      "ci_metric": "octane_score",
      "median_latency_ms": 63985.94270925969,
      "p95_latency_ms": 117054.22423314302,
      "input_tokens": 366,
      "output_tokens": 21,
      "reasoning_tokens": 13960,
      "undefined_reason": null
    },
    {
      "snapshot_id": "db493a14-59b2-4c70-bcec-2cf3993d9e06",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0089715,
      "cost_per_success_usd": 0.0029905,
      "octane_score": 3.270690519979936,
      "ci_low": 2.111653688795038,
      "ci_high": 5.273432155074116,
      "ci_metric": "octane_score",
      "median_latency_ms": 10769.282749854028,
      "p95_latency_ms": 20572.970149479803,
      "input_tokens": 255,
      "output_tokens": 45,
      "reasoning_tokens": 2903,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5843acd4-ffeb-4675-a83c-9bb5b49ff0d7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0007695,
      "cost_per_success_usd": 0.0002565,
      "octane_score": 6.509421702404159,
      "ci_low": 5.708262108262109,
      "ci_high": 7.894405043341215,
      "ci_metric": "octane_score",
      "median_latency_ms": 1593.3666247874498,
      "p95_latency_ms": 1717.943311110139,
      "input_tokens": 261,
      "output_tokens": 3,
      "reasoning_tokens": 210,
      "undefined_reason": null
    },
    {
      "snapshot_id": "15cdd760-2bd9-4912-b8c8-4443b55b2992",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004872,
      "cost_per_success_usd": 0.0016239999999999998,
      "octane_score": 1.9055829228243024,
      "ci_low": 1.472906403940887,
      "ci_high": 2.4175538206766025,
      "ci_metric": "octane_score",
      "median_latency_ms": 7250.044084154069,
      "p95_latency_ms": 9517.61407116428,
      "input_tokens": 408,
      "output_tokens": 131,
      "reasoning_tokens": 1425,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f9ef7ac9-3515-4310-9435-94a78cf2d80d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.015225,
      "cost_per_success_usd": 0.005075,
      "octane_score": 0.5802955665024631,
      "ci_low": 0.22402883863974388,
      "ci_high": 3.352761904761904,
      "ci_metric": "octane_score",
      "median_latency_ms": 5429.977458901703,
      "p95_latency_ms": 56336.9877712801,
      "input_tokens": 444,
      "output_tokens": 102,
      "reasoning_tokens": 4899,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1f7c00d1-abbd-4568-9ce9-d13db32a5c82",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007143,
      "cost_per_success_usd": 0.002381,
      "octane_score": 0.9833403331933361,
      "ci_low": 0.5623827392120075,
      "ci_high": 1.9918628305725075,
      "ci_metric": "octane_score",
      "median_latency_ms": 8364.776874892414,
      "p95_latency_ms": 20892.14833769947,
      "input_tokens": 492,
      "output_tokens": 9,
      "reasoning_tokens": 2290,
      "undefined_reason": null
    },
    {
      "snapshot_id": "046d29db-3a89-458c-ad65-99b6d69312ae",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-statement-field-extraction",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002249,
      "cost_per_success_usd": 0.0007496666666666667,
      "octane_score": 4.128056914184081,
      "ci_low": 3.082272468174045,
      "ci_high": 5.5587318087318085,
      "ci_metric": "octane_score",
      "median_latency_ms": 6125.435541383922,
      "p95_latency_ms": 6460.098366905004,
      "input_tokens": 453,
      "output_tokens": 138,
      "reasoning_tokens": 671,
      "undefined_reason": null
    },
    {
      "snapshot_id": "07fab4c2-ed7a-415a-81b0-934fe84c5aad",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-day-count-convention",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0094155,
      "cost_per_success_usd": 0.0031385000000000002,
      "octane_score": 0.968296957145133,
      "ci_low": 0.5679057559665508,
      "ci_high": 1.85757946210269,
      "ci_metric": "octane_score",
      "median_latency_ms": 13957.696000114083,
      "p95_latency_ms": 30189.849437493827,
      "input_tokens": 411,
      "output_tokens": 21,
      "reasoning_tokens": 3663,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1cd62633-edad-4343-a173-b2f4390e9f32",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-payment-waterfall",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0021595,
      "cost_per_success_usd": 0.0007198333333333333,
      "octane_score": 4.091224820560314,
      "ci_low": 4.042042042042042,
      "ci_high": 4.155128501968049,
      "ci_metric": "octane_score",
      "median_latency_ms": 5349.093708209693,
      "p95_latency_ms": 5463.332133367658,
      "input_tokens": 489,
      "output_tokens": 93,
      "reasoning_tokens": 673,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1ab28f21-98e3-4a90-8a36-b02ac5973208",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-kyc-risk-tier",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0014554999999999998,
      "cost_per_success_usd": 0.00048516666666666657,
      "octane_score": 3.441429062177947,
      "ci_low": 1.5962396430847678,
      "ci_high": 9.486742424242424,
      "ci_metric": "octane_score",
      "median_latency_ms": 2461.314250715077,
      "p95_latency_ms": 8262.690737377852,
      "input_tokens": 306,
      "output_tokens": 3,
      "reasoning_tokens": 518,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b8903619-b44c-4e97-8937-8ed024b257c8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-fx-missing-rate-refusal",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003386,
      "cost_per_success_usd": 0.0011286666666666667,
      "octane_score": 2.0744240992321323,
      "ci_low": 1.5159166936441133,
      "ci_high": 2.7445903458479743,
      "ci_metric": "octane_score",
      "median_latency_ms": 10587.457166053355,
      "p95_latency_ms": 21206.16757906973,
      "input_tokens": 537,
      "output_tokens": 9,
      "reasoning_tokens": 1238,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0b6788d0-4d33-4dbf-ad4f-38c7eb25f69c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "banking",
      "task_id": "banking-effective-annual-rate",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.013695,
      "cost_per_success_usd": 0.0045650000000000005,
      "octane_score": 2.1426067907995616,
      "ci_low": 1.2669922132770226,
      "ci_high": 5.093810178817057,
      "ci_metric": "octane_score",
      "median_latency_ms": 27685.13554148376,
      "p95_latency_ms": 54369.857216533266,
      "input_tokens": 300,
      "output_tokens": 45,
      "reasoning_tokens": 5373,
      "undefined_reason": null
    },
    {
      "snapshot_id": "85ce138c-9e1b-4fed-a1af-b6a6635ce913",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001083,
      "cost_per_success_usd": 0.000361,
      "octane_score": 7.866112650046168,
      "ci_low": 4.6011080332409975,
      "ci_high": 9.592797783933518,
      "ci_metric": "octane_score",
      "median_latency_ms": 748.3745841309428,
      "p95_latency_ms": 769.4200711324811,
      "input_tokens": 153,
      "output_tokens": 186,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9583ad50-a91f-4818-be56-3dc896b79c3a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000516,
      "cost_per_success_usd": 0.00017199999999999998,
      "octane_score": 21.540697674418613,
      "ci_low": 19.069767441860467,
      "ci_high": 26.383720930232567,
      "ci_metric": "octane_score",
      "median_latency_ms": 516.7177505791187,
      "p95_latency_ms": 621.6102749109268,
      "input_tokens": 291,
      "output_tokens": 45,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7c0c06d7-c715-40c1-a7a7-8fe8de90de36",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003372,
      "cost_per_success_usd": 0.001124,
      "octane_score": 5.7313167259786475,
      "ci_low": 4.884341637010677,
      "ci_high": 7.047153024911033,
      "ci_metric": "octane_score",
      "median_latency_ms": 2175.5019165575504,
      "p95_latency_ms": 2325.0600916333497,
      "input_tokens": 282,
      "output_tokens": 618,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ebdb1a2f-f372-4535-92bb-f0f6721d7de8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.005373,
      "cost_per_success_usd": 0.001791,
      "octane_score": 7.219616601526149,
      "ci_low": 5.314907872696817,
      "ci_high": 9.491345616973758,
      "ci_metric": "octane_score",
      "median_latency_ms": 3174.480624496937,
      "p95_latency_ms": 3189.1551626846194,
      "input_tokens": 408,
      "output_tokens": 993,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8f3c78ad-2dd0-4a15-a069-e62d8c86ca95",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.005658,
      "cost_per_success_usd": 0.0018859999999999999,
      "octane_score": 9.76263697419583,
      "ci_low": 8.314422057264052,
      "ci_high": 12.1813361611877,
      "ci_metric": "octane_score",
      "median_latency_ms": 3561.7909152060747,
      "p95_latency_ms": 3566.3203917443752,
      "input_tokens": 438,
      "output_tokens": 1044,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a0c67602-7273-4df3-aada-888786fbafb9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004281,
      "cost_per_success_usd": 0.001427,
      "octane_score": 8.019621583742117,
      "ci_low": 7.569726699369307,
      "ci_high": 8.597757533286615,
      "ci_metric": "octane_score",
      "median_latency_ms": 2529.7952089458704,
      "p95_latency_ms": 2707.633295841515,
      "input_tokens": 426,
      "output_tokens": 771,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7684de3e-014a-4da4-ae1b-1d00fad9ca8a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0036990000000000005,
      "cost_per_success_usd": 0.0012330000000000002,
      "octane_score": 16.585563665855634,
      "ci_low": 8.317923763179234,
      "ci_high": 38.451743714517434,
      "ci_metric": "octane_score",
      "median_latency_ms": 1878.7419172003865,
      "p95_latency_ms": 1902.2527297958732,
      "input_tokens": 354,
      "output_tokens": 669,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5d1564fe-1459-45f2-8219-f8b1bb4c1dcd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.005265000000000001,
      "cost_per_success_usd": 0.0017550000000000005,
      "octane_score": 1.6180436847103508,
      "ci_low": 0.9464387464387463,
      "ci_high": 1.9732193732193726,
      "ci_metric": "octane_score",
      "median_latency_ms": 1971.990916877985,
      "p95_latency_ms": 2339.518179744482,
      "input_tokens": 153,
      "output_tokens": 180,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f6f0b22f-397e-4295-aec9-97524355fc20",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.030440000000000002,
      "cost_per_success_usd": 0.010146666666666667,
      "octane_score": 1.8146189224704339,
      "ci_low": 1.4348047284153824,
      "ci_high": 2.2767158201347173,
      "ci_metric": "octane_score",
      "median_latency_ms": 6281.365083530545,
      "p95_latency_ms": 8349.345332197845,
      "input_tokens": 438,
      "output_tokens": 1130,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9cf15e41-f933-4c31-a069-508a75d88f34",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.021679999999999998,
      "cost_per_success_usd": 0.007226666666666665,
      "octane_score": 1.5835793357933585,
      "ci_low": 1.4717868338557993,
      "ci_high": 1.7022824536376604,
      "ci_metric": "octane_score",
      "median_latency_ms": 5051.212999969721,
      "p95_latency_ms": 5183.738449588418,
      "input_tokens": 426,
      "output_tokens": 782,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ebaf6b93-7fa4-42b7-95b3-ed6d64d8ac65",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.024915,
      "cost_per_success_usd": 0.008305,
      "octane_score": 1.5569335741521173,
      "ci_low": 1.1461770018061408,
      "ci_high": 1.9941985311606871,
      "ci_metric": "octane_score",
      "median_latency_ms": 4874.870624393225,
      "p95_latency_ms": 4970.364824123681,
      "input_tokens": 408,
      "output_tokens": 915,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1083399f-dcbf-46c4-8074-3ae5ee249bb0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.018145,
      "cost_per_success_usd": 0.0060483333333333335,
      "octane_score": 3.3810967208597407,
      "ci_low": 1.726599326599326,
      "ci_high": 7.981649831649832,
      "ci_metric": "octane_score",
      "median_latency_ms": 4231.3127079978585,
      "p95_latency_ms": 4567.958083283156,
      "input_tokens": 354,
      "output_tokens": 655,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "68c79be1-5230-46d8-88b8-a357c689166b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.011609999999999999,
      "cost_per_success_usd": 0.00387,
      "octane_score": 1.6645994832041342,
      "ci_low": 1.4186046511627908,
      "ci_high": 2.0467700258397934,
      "ci_metric": "octane_score",
      "median_latency_ms": 2787.052541039884,
      "p95_latency_ms": 2811.4437418058515,
      "input_tokens": 282,
      "output_tokens": 408,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "45b6ae83-539f-4897-a49c-d1fff6702ee4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0025800000000000003,
      "cost_per_success_usd": 0.0008600000000000001,
      "octane_score": 4.308139534883721,
      "ci_low": 3.8139534883720927,
      "ci_high": 5.276744186046512,
      "ci_metric": "octane_score",
      "median_latency_ms": 1807.7657502144575,
      "p95_latency_ms": 2563.1213879212733,
      "input_tokens": 291,
      "output_tokens": 45,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "98c19301-1be9-47e6-b3aa-95056578c8c8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.011418000000000001,
      "cost_per_success_usd": 0.003806000000000001,
      "octane_score": 3.006831318970047,
      "ci_low": 2.737448561434121,
      "ci_high": 3.304624757684852,
      "ci_metric": "octane_score",
      "median_latency_ms": 4331.765082664788,
      "p95_latency_ms": 4537.786370981485,
      "input_tokens": 426,
      "output_tokens": 676,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "37d91e86-fcce-4a86-b46a-e011d3b9101b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.006867000000000001,
      "cost_per_success_usd": 0.0022890000000000002,
      "octane_score": 8.93403232852774,
      "ci_low": 4.570409982174687,
      "ci_high": 21.12789661319073,
      "ci_metric": "octane_score",
      "median_latency_ms": 2496.596292592585,
      "p95_latency_ms": 2680.7939666323364,
      "input_tokens": 354,
      "output_tokens": 387,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e070eb0a-48c7-4189-9b87-3f4f192e9c3b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.007026,
      "cost_per_success_usd": 0.002342,
      "octane_score": 2.7506404782237404,
      "ci_low": 2.354202401372213,
      "ci_high": 3.382152006831768,
      "ci_metric": "octane_score",
      "median_latency_ms": 2286.236292682588,
      "p95_latency_ms": 2775.6107295863326,
      "input_tokens": 282,
      "output_tokens": 412,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "309d4417-d118-41a9-ad66-030f730cf6fc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001548,
      "cost_per_success_usd": 0.000516,
      "octane_score": 7.1802325581395365,
      "ci_low": 6.356589147286822,
      "ci_high": 8.794573643410855,
      "ci_metric": "octane_score",
      "median_latency_ms": 1242.0498337596655,
      "p95_latency_ms": 2291.3323456421494,
      "input_tokens": 291,
      "output_tokens": 45,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "43df241b-8ad3-426d-8b8b-6fc5930a83f5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0031589999999999995,
      "cost_per_success_usd": 0.001053,
      "octane_score": 2.6967394745172526,
      "ci_low": 1.5773979107312444,
      "ci_high": 3.2886989553656223,
      "ci_metric": "octane_score",
      "median_latency_ms": 1619.4711672142148,
      "p95_latency_ms": 2463.489028532058,
      "input_tokens": 153,
      "output_tokens": 180,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b91c9e85-a324-4236-ba10-ed01d0418bb6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.014079000000000001,
      "cost_per_success_usd": 0.004693,
      "octane_score": 2.755238298174586,
      "ci_low": 2.0457769181173435,
      "ci_high": 3.62220328148306,
      "ci_metric": "octane_score",
      "median_latency_ms": 4372.727459296584,
      "p95_latency_ms": 4422.182383667678,
      "input_tokens": 408,
      "output_tokens": 857,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8c797cba-136e-4596-a474-a87ec368370d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.019899,
      "cost_per_success_usd": 0.006633,
      "octane_score": 2.77586813407709,
      "ci_low": 2.3748296228986825,
      "ci_high": 3.447996398018911,
      "ci_metric": "octane_score",
      "median_latency_ms": 6473.550166934729,
      "p95_latency_ms": 6526.992053352296,
      "input_tokens": 438,
      "output_tokens": 1239,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7ad250f7-e490-40e3-bae7-b1fa9751bb52",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005331800000000001,
      "cost_per_success_usd": 0.00017772666666666668,
      "octane_score": 64.39101241606961,
      "ci_low": 60.39140022050716,
      "ci_high": 69.03297197944408,
      "ci_metric": "octane_score",
      "median_latency_ms": 9335.692624561489,
      "p95_latency_ms": 10032.91024947539,
      "input_tokens": 405,
      "output_tokens": 1126,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "18246893-469d-4b96-92f9-1dae008b9cd5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0004983,
      "cost_per_success_usd": 0.0001661,
      "octane_score": 110.85089303632351,
      "ci_low": 93.5508889154039,
      "ci_high": 134.61853978671044,
      "ci_metric": "octane_score",
      "median_latency_ms": 6079.76037543267,
      "p95_latency_ms": 8220.343924220651,
      "input_tokens": 417,
      "output_tokens": 1026,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4468fb4b-4e44-49fc-9093-873e71fcfa96",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00017586000000000001,
      "cost_per_success_usd": 5.8620000000000005e-05,
      "octane_score": 348.8570453770044,
      "ci_low": 174.9573524394404,
      "ci_high": 808.7853974752645,
      "ci_metric": "octane_score",
      "median_latency_ms": 2682.7696254476905,
      "p95_latency_ms": 2905.9842507354915,
      "input_tokens": 330,
      "output_tokens": 237,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b123cd89-78d6-4fc8-88fb-63fc3836836e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0002211,
      "cost_per_success_usd": 7.37e-05,
      "octane_score": 87.40841248303934,
      "ci_low": 74.49118046132972,
      "ci_high": 107.4762550881954,
      "ci_metric": "octane_score",
      "median_latency_ms": 3945.226832292974,
      "p95_latency_ms": 4293.9199461601675,
      "input_tokens": 276,
      "output_tokens": 393,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "45be0c49-cc1c-43e0-ac44-ef7f12144cfa",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 8.268e-05,
      "cost_per_success_usd": 2.756e-05,
      "octane_score": 134.43396226415098,
      "ci_low": 119.01306240928882,
      "ci_high": 164.65892597968073,
      "ci_metric": "octane_score",
      "median_latency_ms": 1641.214874573052,
      "p95_latency_ms": 1697.1850874833763,
      "input_tokens": 261,
      "output_tokens": 39,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "adb931bf-5100-4a35-82a6-098235fde8dd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 9.258e-05,
      "cost_per_success_usd": 3.086e-05,
      "octane_score": 92.01771440915964,
      "ci_low": 53.823720025923535,
      "ci_high": 112.21646143875567,
      "ci_metric": "octane_score",
      "median_latency_ms": 2534.105334430933,
      "p95_latency_ms": 3551.354483049363,
      "input_tokens": 150,
      "output_tokens": 141,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "783cd6bb-17d8-42ff-8dde-f604d64cf891",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00037569999999999997,
      "cost_per_success_usd": 0.00012523333333333333,
      "octane_score": 103.24993345754592,
      "ci_low": 76.01011445302103,
      "ci_high": 132.96264367816096,
      "ci_metric": "octane_score",
      "median_latency_ms": 4122.267374768853,
      "p95_latency_ms": 4816.242488101125,
      "input_tokens": 375,
      "output_tokens": 754,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "03694719-bc92-43c9-a4fc-b966c20032a6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 6.2604e-05,
      "cost_per_success_usd": 2.0868e-05,
      "octane_score": 177.5445658424382,
      "ci_low": 133.1307401404646,
      "ci_high": 279.3214537400583,
      "ci_metric": "octane_score",
      "median_latency_ms": 1331.457500346005,
      "p95_latency_ms": 1572.026937454939,
      "input_tokens": 480,
      "output_tokens": 140,
      "reasoning_tokens": 51,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3dd041ed-1784-4ec4-8953-568819f049c9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.00642548,
      "cost_per_success_usd": 0.00642548,
      "octane_score": 1.7810342573628744,
      "ci_low": 1.7032556982493239,
      "ci_high": 22.875189893659552,
      "ci_metric": "octane_score",
      "median_latency_ms": 104939.97404165566,
      "p95_latency_ms": 142206.4800299704,
      "input_tokens": 466,
      "output_tokens": 7906,
      "reasoning_tokens": 27584,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2fe39a6f-a9ce-45e2-98c9-9a05f805398e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00366422,
      "cost_per_success_usd": 0.00183211,
      "octane_score": 11.161993548422311,
      "ci_low": 1.860786000681785,
      "ci_high": 152.4077407740774,
      "ci_metric": "octane_score",
      "median_latency_ms": 22053.384041413665,
      "p95_latency_ms": 216276.9526164047,
      "input_tokens": 391,
      "output_tokens": 1079,
      "reasoning_tokens": 19104,
      "undefined_reason": null
    },
    {
      "snapshot_id": "76606191-133a-46ec-8665-00fbb948afd7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00308158,
      "cost_per_success_usd": 0.00154079,
      "octane_score": 4.180972098728574,
      "ci_low": 0.941643202222267,
      "ci_high": 124.1129492298916,
      "ci_metric": "octane_score",
      "median_latency_ms": 13982.87916649133,
      "p95_latency_ms": 78031.5960658714,
      "input_tokens": 416,
      "output_tokens": 181,
      "reasoning_tokens": 16754,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d081ff39-8837-42ef-a32f-6ffff1cc61c8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016334,
      "cost_per_success_usd": 0.0005444666666666666,
      "octane_score": 33.81719113505572,
      "ci_low": 20.27924018474055,
      "ci_high": 76.27389876189675,
      "ci_metric": "octane_score",
      "median_latency_ms": 98637.67349999398,
      "p95_latency_ms": 125685.30566217378,
      "input_tokens": 478,
      "output_tokens": 1694,
      "reasoning_tokens": 7168,
      "undefined_reason": null
    },
    {
      "snapshot_id": "111c4871-06f9-4f18-9625-bb5ba21cdd50",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.03943304,
      "cost_per_success_usd": 0.01971652,
      "octane_score": 0.6558121480531723,
      "ci_low": 0.13327660213492848,
      "ci_high": 87.61575642589331,
      "ci_metric": "octane_score",
      "median_latency_ms": 64147.19483349472,
      "p95_latency_ms": 3246644.931046478,
      "input_tokens": 442,
      "output_tokens": 1545,
      "reasoning_tokens": 217331,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f6f1dc36-3638-444e-9b27-ede2369fbb06",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00012652,
      "cost_per_success_usd": 4.217333333333334e-05,
      "octane_score": 67.33322794815048,
      "ci_low": 37.7715704531347,
      "ci_high": 135.65758137073306,
      "ci_metric": "octane_score",
      "median_latency_ms": 2072.165166027844,
      "p95_latency_ms": 6241.473741549998,
      "input_tokens": 290,
      "output_tokens": 153,
      "reasoning_tokens": 421,
      "undefined_reason": null
    },
    {
      "snapshot_id": "33ae7607-cf58-42bc-9cc9-3b1c9bee241b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00147813,
      "cost_per_success_usd": 0.0004927099999999999,
      "octane_score": 5.763363168327549,
      "ci_low": 2.7909231930991893,
      "ci_high": 18.089000192641112,
      "ci_metric": "octane_score",
      "median_latency_ms": 6535.479791462421,
      "p95_latency_ms": 17300.681442022324,
      "input_tokens": 290,
      "output_tokens": 120,
      "reasoning_tokens": 1434,
      "undefined_reason": null
    },
    {
      "snapshot_id": "786668fe-7384-434b-9303-876ff019ba9a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.02954607,
      "cost_per_success_usd": 0.014773035,
      "octane_score": 0.8752658701027469,
      "ci_low": 0.2773576797941861,
      "ci_high": 2.1559470376287804,
      "ci_metric": "octane_score",
      "median_latency_ms": 173642.38583296537,
      "p95_latency_ms": 179232.84941967577,
      "input_tokens": 594,
      "output_tokens": 304,
      "reasoning_tokens": 33360,
      "undefined_reason": null
    },
    {
      "snapshot_id": "51c11fe5-3b8c-429c-a660-4a673b9a0c6a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0018879,
      "cost_per_success_usd": 0.0006293000000000001,
      "octane_score": 32.49642459876053,
      "ci_low": 14.966901252138035,
      "ci_high": 95.74594840208007,
      "ci_metric": "octane_score",
      "median_latency_ms": 9636.249667033553,
      "p95_latency_ms": 11338.054942153394,
      "input_tokens": 470,
      "output_tokens": 229,
      "reasoning_tokens": 1706,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ce88eb70-05da-409c-bef5-ee90f285cb00",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.24026094,
      "cost_per_success_usd": 0.12013047,
      "octane_score": 0.09526309187003099,
      "ci_low": 0.03228823929408661,
      "ci_high": 1.3862391177685478,
      "ci_metric": "octane_score",
      "median_latency_ms": 749616.1389583722,
      "p95_latency_ms": 2847643.538570776,
      "input_tokens": 624,
      "output_tokens": 4508,
      "reasoning_tokens": 271342,
      "undefined_reason": null
    },
    {
      "snapshot_id": "76e26aa8-cda3-4fba-8112-afc846132ad8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.043004534999999997,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 251420.49174942076,
      "p95_latency_ms": 289975.3423749469,
      "input_tokens": 557,
      "output_tokens": 610,
      "reasoning_tokens": 48542,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "5f62c55c-e811-46bd-98bb-2f9ccee2a7f0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00110838,
      "cost_per_success_usd": 0.00055419,
      "octane_score": 11.62417221530522,
      "ci_low": 8.837105906659868,
      "ci_high": 15.3943034638421,
      "ci_metric": "octane_score",
      "median_latency_ms": 12095.916625112295,
      "p95_latency_ms": 97308.35957489906,
      "input_tokens": 330,
      "output_tokens": 157,
      "reasoning_tokens": 952,
      "undefined_reason": null
    },
    {
      "snapshot_id": "34848cea-bacd-42c9-adef-424259e330ba",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00051678,
      "cost_per_success_usd": 0.00017226,
      "octane_score": 21.508185301288755,
      "ci_low": 18.33314734218651,
      "ci_high": 25.950706238920347,
      "ci_metric": "octane_score",
      "median_latency_ms": 3906.971584074199,
      "p95_latency_ms": 4130.902909021825,
      "input_tokens": 480,
      "output_tokens": 38,
      "reasoning_tokens": 316,
      "undefined_reason": null
    },
    {
      "snapshot_id": "afd587c1-b8d7-4d2f-a223-00b4d27088aa",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0011247000000000002,
      "cost_per_success_usd": 0.00037490000000000006,
      "octane_score": 17.183248866364362,
      "ci_low": 11.654570130406567,
      "ci_high": 25.91044420626414,
      "ci_metric": "octane_score",
      "median_latency_ms": 1035.386917181313,
      "p95_latency_ms": 1712.2631173580883,
      "input_tokens": 249,
      "output_tokens": 420,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6f79415b-1f48-4722-b136-e68f77d17d46",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0055965,
      "cost_per_success_usd": 0.0018655,
      "octane_score": 6.134548378450818,
      "ci_low": 5.790404717233986,
      "ci_high": 6.576789064593943,
      "ci_metric": "octane_score",
      "median_latency_ms": 4347.683458589017,
      "p95_latency_ms": 4578.95672051236,
      "input_tokens": 405,
      "output_tokens": 2190,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8ca69a1e-d539-4b92-bf38-231e53a45b48",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0061224,
      "cost_per_success_usd": 0.0030612,
      "octane_score": 6.014743673504944,
      "ci_low": 2.237529868578256,
      "ci_high": 12.255017246785828,
      "ci_metric": "octane_score",
      "median_latency_ms": 3477.4632500484586,
      "p95_latency_ms": 6336.750312708317,
      "input_tokens": 408,
      "output_tokens": 2400,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "58909161-02a6-4d46-8201-afb680e3f16f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0032199000000000004,
      "cost_per_success_usd": 0.0010733000000000001,
      "octane_score": 19.053386751141336,
      "ci_low": 9.555576260132298,
      "ci_high": 49.47406866325785,
      "ci_metric": "octane_score",
      "median_latency_ms": 2182.1861676871777,
      "p95_latency_ms": 2771.3257921859617,
      "input_tokens": 333,
      "output_tokens": 1248,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "37ad0d39-040d-476a-9192-fe534c65a710",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0021691,
      "cost_per_success_usd": 0.0007230333333333334,
      "octane_score": 17.88345396708312,
      "ci_low": 13.287887953096643,
      "ci_high": 23.206249741382813,
      "ci_metric": "octane_score",
      "median_latency_ms": 1682.316416874528,
      "p95_latency_ms": 2216.969854477793,
      "input_tokens": 372,
      "output_tokens": 823,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "75cc1427-8c18-40a5-9656-dd108bb7a78b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0004644,
      "cost_per_success_usd": 0.0001548,
      "octane_score": 18.344099913867357,
      "ci_low": 10.72997416020672,
      "ci_high": 22.370801033591732,
      "ci_metric": "octane_score",
      "median_latency_ms": 848.1608340516686,
      "p95_latency_ms": 895.9265710785985,
      "input_tokens": 123,
      "output_tokens": 171,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2322f79e-8dfc-43f9-bba1-574cb3355696",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001692,
      "cost_per_success_usd": 5.6399999999999995e-05,
      "octane_score": 65.69148936170215,
      "ci_low": 58.15602836879433,
      "ci_high": 80.46099290780145,
      "ci_metric": "octane_score",
      "median_latency_ms": 602.4166652932763,
      "p95_latency_ms": 707.1933533996344,
      "input_tokens": 264,
      "output_tokens": 36,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f958fce3-3440-448a-befb-928d58343d0d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.020748,
      "cost_per_success_usd": 0.006915999999999999,
      "octane_score": 0.9314632735685368,
      "ci_low": 0.779901214051609,
      "ci_high": 1.145315211104685,
      "ci_metric": "octane_score",
      "median_latency_ms": 4891.056875698268,
      "p95_latency_ms": 5058.264462184161,
      "input_tokens": 252,
      "output_tokens": 382,
      "reasoning_tokens": 2334,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b07ed673-fd3d-49a5-8fd7-01bf8bd0ac13",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0351405,
      "cost_per_success_usd": 0.0117135,
      "octane_score": 0.9769923592436079,
      "ci_low": 0.8910850821538963,
      "ci_high": 1.0718584388754155,
      "ci_metric": "octane_score",
      "median_latency_ms": 8097.617000341415,
      "p95_latency_ms": 8441.077250149101,
      "input_tokens": 407,
      "output_tokens": 432,
      "reasoning_tokens": 4172,
      "undefined_reason": null
    },
    {
      "snapshot_id": "115759ce-7473-4177-9053-96a35e05e36f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0081165,
      "cost_per_success_usd": 0.0027055,
      "octane_score": 1.369432637220477,
      "ci_low": 1.0544501497646555,
      "ci_high": 1.7970951343500365,
      "ci_metric": "octane_score",
      "median_latency_ms": 2421.765082515776,
      "p95_latency_ms": 2856.0930827632546,
      "input_tokens": 266,
      "output_tokens": 36,
      "reasoning_tokens": 993,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b6a4bc6b-165e-43e9-992f-b8efdda73139",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.093195,
      "cost_per_success_usd": 0.031065,
      "octane_score": 0.5927034712162671,
      "ci_low": 0.47457560009898536,
      "ci_high": 0.7395461129888943,
      "ci_metric": "octane_score",
      "median_latency_ms": 18717.768625356257,
      "p95_latency_ms": 22032.485875487324,
      "input_tokens": 410,
      "output_tokens": 1019,
      "reasoning_tokens": 11325,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9af3094d-ce7b-424e-a536-3fa08841e9f4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0435975,
      "cost_per_success_usd": 0.02179875,
      "octane_score": 0.9381271861918689,
      "ci_low": 0.2421035957547917,
      "ci_high": 3.569433465085639,
      "ci_metric": "octane_score",
      "median_latency_ms": 11559.030207805336,
      "p95_latency_ms": 11963.764745555818,
      "input_tokens": 335,
      "output_tokens": 337,
      "reasoning_tokens": 5409,
      "undefined_reason": null
    },
    {
      "snapshot_id": "86086833-45b3-4f8c-ba2a-5726ad10d1a8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.010029,
      "cost_per_success_usd": 0.003343,
      "octane_score": 0.8494366337620899,
      "ci_low": 0.5361523563589413,
      "ci_high": 1.2016532774478987,
      "ci_metric": "octane_score",
      "median_latency_ms": 2781.331917271018,
      "p95_latency_ms": 3188.595903851092,
      "input_tokens": 126,
      "output_tokens": 64,
      "reasoning_tokens": 1248,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9a11b284-409e-4a37-a2f0-88d9c97d8b3e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0382725,
      "cost_per_success_usd": 0.0127575,
      "octane_score": 1.0135475863870926,
      "ci_low": 0.7536817102137768,
      "ci_high": 1.3259115982520242,
      "ci_metric": "octane_score",
      "median_latency_ms": 7560.6385404244065,
      "p95_latency_ms": 8773.135978728533,
      "input_tokens": 375,
      "output_tokens": 600,
      "reasoning_tokens": 4428,
      "undefined_reason": null
    },
    {
      "snapshot_id": "50a6b558-a541-44ed-870c-1846cdb1067f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.006905625,
      "cost_per_success_usd": 0.002301875,
      "octane_score": 4.971599239750204,
      "ci_low": 4.337929400615969,
      "ci_high": 5.7335735735735724,
      "ci_metric": "octane_score",
      "median_latency_ms": 7216.268541291356,
      "p95_latency_ms": 8007.706941105425,
      "input_tokens": 405,
      "output_tokens": 440,
      "reasoning_tokens": 3162,
      "undefined_reason": null
    },
    {
      "snapshot_id": "508ae235-7ca8-497a-a1a3-38a153452dd2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003494625,
      "cost_per_success_usd": 0.0011648750000000001,
      "octane_score": 5.5302071037664975,
      "ci_low": 4.602444921170022,
      "ci_high": 6.696522006771314,
      "ci_metric": "octane_score",
      "median_latency_ms": 4170.798582956195,
      "p95_latency_ms": 4195.453558862209,
      "input_tokens": 249,
      "output_tokens": 376,
      "reasoning_tokens": 1438,
      "undefined_reason": null
    },
    {
      "snapshot_id": "76ff2d46-d140-4ee6-86a2-2f17e965fa49",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0070151250000000005,
      "cost_per_success_usd": 0.002338375,
      "octane_score": 5.529623492097432,
      "ci_low": 4.050844168445566,
      "ci_high": 7.343728921160181,
      "ci_metric": "octane_score",
      "median_latency_ms": 7328.102916479111,
      "p95_latency_ms": 7589.710341300815,
      "input_tokens": 372,
      "output_tokens": 656,
      "reasoning_tokens": 3011,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8a0d8095-7dd0-4b87-8541-fa3b761ff237",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0069967499999999995,
      "cost_per_success_usd": 0.0023322499999999997,
      "octane_score": 8.768356737056491,
      "ci_low": 4.397470254046521,
      "ci_high": 21.90769941662335,
      "ci_metric": "octane_score",
      "median_latency_ms": 8359.564208425581,
      "p95_latency_ms": 10080.856808274983,
      "input_tokens": 333,
      "output_tokens": 379,
      "reasoning_tokens": 3286,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d799fc16-bcfb-47e7-a69a-f9f3b6ee17d6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001279875,
      "cost_per_success_usd": 0.00042662500000000003,
      "octane_score": 6.656118761597812,
      "ci_low": 4.26170622193714,
      "ci_high": 9.05946894886987,
      "ci_metric": "octane_score",
      "median_latency_ms": 2494.3889593705535,
      "p95_latency_ms": 3123.896732274443,
      "input_tokens": 123,
      "output_tokens": 69,
      "reasoning_tokens": 589,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5bff8356-1357-48cb-852c-71a3e5dde8e2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.010842375000000001,
      "cost_per_success_usd": 0.0036141250000000006,
      "octane_score": 5.094548011851646,
      "ci_low": 4.2176843818383745,
      "ci_high": 6.351233671988389,
      "ci_metric": "octane_score",
      "median_latency_ms": 11772.113291546702,
      "p95_latency_ms": 12580.728267040104,
      "input_tokens": 408,
      "output_tokens": 970,
      "reasoning_tokens": 4731,
      "undefined_reason": null
    },
    {
      "snapshot_id": "90e7f388-7c5c-4879-aa1e-f03c9dfdd300",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001239,
      "cost_per_success_usd": 0.000413,
      "octane_score": 8.970944309927363,
      "ci_low": 7.850188829258599,
      "ci_high": 10.921780986762938,
      "ci_metric": "octane_score",
      "median_latency_ms": 2360.9696673229337,
      "p95_latency_ms": 3698.322304151952,
      "input_tokens": 264,
      "output_tokens": 36,
      "reasoning_tokens": 572,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ffc91057-d226-436e-b919-5802a9f4d3de",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.007930020000000001,
      "cost_per_success_usd": 0.003965010000000001,
      "octane_score": 5.1576162481305206,
      "ci_low": 0.8384644755263169,
      "ci_high": 67.81367113699014,
      "ci_metric": "octane_score",
      "median_latency_ms": 38136.57058402896,
      "p95_latency_ms": 159623.67857154456,
      "input_tokens": 312,
      "output_tokens": 226,
      "reasoning_tokens": 19578,
      "undefined_reason": null
    },
    {
      "snapshot_id": "70939e92-22d6-40f9-8a69-613078f0fd1f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0170636,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 241471.87825012952,
      "p95_latency_ms": 249314.3571623601,
      "input_tokens": 260,
      "output_tokens": 1998,
      "reasoning_tokens": 40622,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "c909cb64-33d7-418e-87cf-f6b0f5712b18",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.01508416,
      "cost_per_success_usd": 0.01508416,
      "octane_score": 1.2206402831402834,
      "ci_low": 1.028865929824923,
      "ci_high": 7.932521081776613,
      "ci_metric": "octane_score",
      "median_latency_ms": 292081.4499575645,
      "p95_latency_ms": 345928.4074706025,
      "input_tokens": 396,
      "output_tokens": 1602,
      "reasoning_tokens": 36049,
      "undefined_reason": null
    },
    {
      "snapshot_id": "80a1212e-8f22-4862-adcc-ed5cb0beebc5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00230552,
      "cost_per_success_usd": 0.00115276,
      "octane_score": 2.463363290421828,
      "ci_low": 0.7305724503380405,
      "ci_high": 6.674556366082053,
      "ci_metric": "octane_score",
      "median_latency_ms": 28593.130249530077,
      "p95_latency_ms": 55758.47033793106,
      "input_tokens": 132,
      "output_tokens": 0,
      "reasoning_tokens": 5744,
      "undefined_reason": null
    },
    {
      "snapshot_id": "43ac02c9-8542-4e06-8182-1f343a6a8133",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.01461782,
      "cost_per_success_usd": 0.01461782,
      "octane_score": 0.8845596219773765,
      "ci_low": 0.7083125777908682,
      "ci_high": 9.37564955938725,
      "ci_metric": "octane_score",
      "median_latency_ms": 211296.9820415601,
      "p95_latency_ms": 309117.97559121624,
      "input_tokens": 357,
      "output_tokens": 3110,
      "reasoning_tokens": 33381,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f04a04ff-95cb-487d-9635-1b4edec29845",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0009169800000000001,
      "cost_per_success_usd": 0.00030566,
      "octane_score": 12.121311260878102,
      "ci_low": 10.571302939630856,
      "ci_high": 14.769250797370308,
      "ci_metric": "octane_score",
      "median_latency_ms": 10535.941707901657,
      "p95_latency_ms": 18789.178670942783,
      "input_tokens": 243,
      "output_tokens": 42,
      "reasoning_tokens": 2234,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b4e6f416-92cf-4441-9aac-38bf832f2ee7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0007021,
      "cost_per_success_usd": 0.00023403333333333335,
      "octane_score": 27.525993448226746,
      "ci_low": 20.08730901153726,
      "ci_high": 38.40600176522507,
      "ci_metric": "octane_score",
      "median_latency_ms": 7676.147458143532,
      "p95_latency_ms": 12160.224358364938,
      "input_tokens": 255,
      "output_tokens": 158,
      "reasoning_tokens": 1559,
      "undefined_reason": null
    },
    {
      "snapshot_id": "471ee086-ee8c-4d82-8223-60d4b0b130b1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00498996,
      "cost_per_success_usd": 0.00166332,
      "octane_score": 2.2274727653127484,
      "ci_low": 1.8440982122144147,
      "ci_high": 2.7111654527446913,
      "ci_metric": "octane_score",
      "median_latency_ms": 21983.21441654116,
      "p95_latency_ms": 27632.25225387141,
      "input_tokens": 243,
      "output_tokens": 733,
      "reasoning_tokens": 1806,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b1990072-3a6e-4c1a-8da1-ac6aec5a1c3e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.09462167999999999,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 185596.40004020184,
      "p95_latency_ms": 282800.6619160995,
      "input_tokens": 410,
      "output_tokens": 1,
      "reasoning_tokens": 49153,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "fd8f3d1c-c0ab-484a-813f-92d642a84ebe",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.07128456,
      "cost_per_success_usd": 0.07128456,
      "octane_score": 0.16053967366846342,
      "ci_low": 0.1536798431525705,
      "ci_high": 1.3725377315947536,
      "ci_metric": "octane_score",
      "median_latency_ms": 2534352.4756655097,
      "p95_latency_ms": 3125422.445829585,
      "input_tokens": 390,
      "output_tokens": 1130,
      "reasoning_tokens": 35942,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8b5e6326-3f67-4063-bd4a-ce80cdb1175b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00921732,
      "cost_per_success_usd": 0.00460866,
      "octane_score": 4.437298477214635,
      "ci_low": 1.6631749182195759,
      "ci_high": 13.149413683311334,
      "ci_metric": "octane_score",
      "median_latency_ms": 21734.157458879054,
      "p95_latency_ms": 27293.367195967585,
      "input_tokens": 319,
      "output_tokens": 327,
      "reasoning_tokens": 4374,
      "undefined_reason": null
    },
    {
      "snapshot_id": "25a9a177-e6b0-4300-b744-00ccdc66e095",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004882559999999999,
      "cost_per_success_usd": 0.0016275199999999997,
      "octane_score": 1.7447814261371086,
      "ci_low": 1.0667129058775178,
      "ci_high": 2.5536916072842444,
      "ci_metric": "octane_score",
      "median_latency_ms": 201374.7037500143,
      "p95_latency_ms": 203075.42073726654,
      "input_tokens": 132,
      "output_tokens": 141,
      "reasoning_tokens": 2383,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4701658b-7677-4b9b-8c3a-7aee49ebde89",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00912936,
      "cost_per_success_usd": 0.00304312,
      "octane_score": 2.116906332974053,
      "ci_low": 1.3056555436650776,
      "ci_high": 3.520618523337022,
      "ci_metric": "octane_score",
      "median_latency_ms": 22413.597082719207,
      "p95_latency_ms": 23175.30210847035,
      "input_tokens": 262,
      "output_tokens": 314,
      "reasoning_tokens": 4359,
      "undefined_reason": null
    },
    {
      "snapshot_id": "52e7dc21-5b19-44cf-90db-a3d8065f3d48",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.09458604,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 261290.08012544364,
      "p95_latency_ms": 354096.65842568496,
      "input_tokens": 357,
      "output_tokens": 1792,
      "reasoning_tokens": 47360,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "9f8bfd10-5d11-4052-bd94-a6cae480e25b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.012919008,
      "cost_per_success_usd": 0.004306336,
      "octane_score": 3.002629923288228,
      "ci_low": 1.992210888721899,
      "ci_high": 4.571643044024436,
      "ci_metric": "octane_score",
      "median_latency_ms": 10060.66033244133,
      "p95_latency_ms": 11264.348720759153,
      "input_tokens": 378,
      "output_tokens": 866,
      "reasoning_tokens": 3325,
      "undefined_reason": null
    },
    {
      "snapshot_id": "19b2b055-1e7d-4735-9943-d94808c59813",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.018417562800000002,
      "cost_per_success_usd": 0.006139187600000001,
      "octane_score": 1.8640902910345987,
      "ci_low": 1.6966248332103957,
      "ci_high": 2.0498371708158705,
      "ci_metric": "octane_score",
      "median_latency_ms": 21082.002541981637,
      "p95_latency_ms": 23774.632891081274,
      "input_tokens": 411,
      "output_tokens": 425,
      "reasoning_tokens": 5602,
      "undefined_reason": null
    },
    {
      "snapshot_id": "833f51d3-c63d-4335-b3c3-c271e06fa880",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0023553012,
      "cost_per_success_usd": 0.0007851004,
      "octane_score": 3.6169471658232077,
      "ci_low": 2.2867788661431723,
      "ci_high": 4.816637368294999,
      "ci_metric": "octane_score",
      "median_latency_ms": 2588.9693330973387,
      "p95_latency_ms": 3286.5607590414584,
      "input_tokens": 153,
      "output_tokens": 103,
      "reasoning_tokens": 646,
      "undefined_reason": null
    },
    {
      "snapshot_id": "462e1c84-b497-43f5-9507-ef48714cfd09",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0031565568000000003,
      "cost_per_success_usd": 0.0010521856000000002,
      "octane_score": 6.122493978248703,
      "ci_low": 4.284938058487591,
      "ci_high": 9.724501442442847,
      "ci_metric": "octane_score",
      "median_latency_ms": 3558.743583969772,
      "p95_latency_ms": 4420.8878967911005,
      "input_tokens": 276,
      "output_tokens": 309,
      "reasoning_tokens": 705,
      "undefined_reason": null
    },
    {
      "snapshot_id": "695b27f5-c7a8-4c51-bf15-1200de66b2b7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.05678559240000001,
      "cost_per_success_usd": 0.028392796200000004,
      "octane_score": 0.7202531182892088,
      "ci_low": 0.11113144045088802,
      "ci_high": 13.645996301353888,
      "ci_metric": "octane_score",
      "median_latency_ms": 10955.793916247785,
      "p95_latency_ms": 141106.57452866435,
      "input_tokens": 333,
      "output_tokens": 234,
      "reasoning_tokens": 18413,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6f1c2041-cf9e-4eff-a300-0e755fa77976",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.085883334,
      "cost_per_success_usd": 0.042941667,
      "octane_score": 0.4287754672712947,
      "ci_low": 0.13194855302982714,
      "ci_high": 1.4647346531925314,
      "ci_metric": "octane_score",
      "median_latency_ms": 74490.08666723967,
      "p95_latency_ms": 191114.67116679993,
      "input_tokens": 417,
      "output_tokens": 733,
      "reasoning_tokens": 27485,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a5d535fc-90cf-4317-8d5f-641b8e0b6ceb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0022404024,
      "cost_per_success_usd": 0.0007468008,
      "octane_score": 4.961162334052134,
      "ci_low": 4.2835617957567145,
      "ci_high": 6.066124770214447,
      "ci_metric": "octane_score",
      "median_latency_ms": 2603.9689583703876,
      "p95_latency_ms": 2735.8046712353826,
      "input_tokens": 264,
      "output_tokens": 42,
      "reasoning_tokens": 671,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3b9f3110-4391-4e9b-972f-e0013728e9d3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0040466,
      "cost_per_success_usd": 0.0013488666666666665,
      "octane_score": 2.1052241387831763,
      "ci_low": 1.3649810989974251,
      "ci_high": 2.8409028351224883,
      "ci_metric": "octane_score",
      "median_latency_ms": 5311.0459158197045,
      "p95_latency_ms": 5626.46467871964,
      "input_tokens": 153,
      "output_tokens": 141,
      "reasoning_tokens": 730,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9085b2bb-6c3a-405c-88a8-df4606d00f25",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.034347960000000004,
      "cost_per_success_usd": 0.01144932,
      "octane_score": 1.6081595529981987,
      "ci_low": 1.3130752098718363,
      "ci_high": 1.9883888852323193,
      "ci_metric": "octane_score",
      "median_latency_ms": 46117.9705420509,
      "p95_latency_ms": 54096.4127917774,
      "input_tokens": 417,
      "output_tokens": 593,
      "reasoning_tokens": 7147,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1737de61-340e-4b68-aa4f-05f408a86b7a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00778648,
      "cost_per_success_usd": 0.0025954933333333335,
      "octane_score": 2.4819944313733546,
      "ci_low": 2.075287678232757,
      "ci_high": 3.0518282972537016,
      "ci_metric": "octane_score",
      "median_latency_ms": 9196.028583683074,
      "p95_latency_ms": 10701.667908299713,
      "input_tokens": 276,
      "output_tokens": 315,
      "reasoning_tokens": 1400,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a5c09f37-9256-40a7-b5ff-fa339b017cd1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.009918080000000001,
      "cost_per_success_usd": 0.003306026666666667,
      "octane_score": 3.911140059366329,
      "ci_low": 2.6527731650614412,
      "ci_high": 5.760085845458258,
      "ci_metric": "octane_score",
      "median_latency_ms": 11556.388707831502,
      "p95_latency_ms": 15756.53945831582,
      "input_tokens": 378,
      "output_tokens": 482,
      "reasoning_tokens": 1685,
      "undefined_reason": null
    },
    {
      "snapshot_id": "79093657-9572-42b7-b170-698b21c13db5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00221688,
      "cost_per_success_usd": 0.00073896,
      "octane_score": 5.013803182851577,
      "ci_low": 4.304667575027077,
      "ci_high": 6.141063115730217,
      "ci_metric": "octane_score",
      "median_latency_ms": 3168.7856251373887,
      "p95_latency_ms": 3631.525162421167,
      "input_tokens": 264,
      "output_tokens": 42,
      "reasoning_tokens": 411,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7748fca8-dd21-482e-9237-058ea8619db8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.012803080000000001,
      "cost_per_success_usd": 0.0042676933333333335,
      "octane_score": 4.791815719342533,
      "ci_low": 2.2758316722389433,
      "ci_high": 13.67199064906702,
      "ci_metric": "octane_score",
      "median_latency_ms": 17819.836957380176,
      "p95_latency_ms": 18583.181033283472,
      "input_tokens": 333,
      "output_tokens": 413,
      "reasoning_tokens": 2424,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e5f2a034-9333-42b4-b572-c973b0dc9c22",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.02771316,
      "cost_per_success_usd": 0.00923772,
      "octane_score": 1.238833824796595,
      "ci_low": 1.043177116128452,
      "ci_high": 1.4399523113086878,
      "ci_metric": "octane_score",
      "median_latency_ms": 35788.36783301085,
      "p95_latency_ms": 41367.32668345793,
      "input_tokens": 411,
      "output_tokens": 436,
      "reasoning_tokens": 5798,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8bae441f-a062-4472-b53c-a25edc9d4558",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0025515,
      "cost_per_success_usd": 0.0008505,
      "octane_score": 13.455614344503234,
      "ci_low": 12.656121851200936,
      "ci_high": 14.425631981187538,
      "ci_metric": "octane_score",
      "median_latency_ms": 1354.6108333393931,
      "p95_latency_ms": 1355.5736080743372,
      "input_tokens": 390,
      "output_tokens": 502,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "01b89c5c-4f7e-49d2-97ae-0eddc1577e57",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002781,
      "cost_per_success_usd": 0.000927,
      "octane_score": 13.948579647608774,
      "ci_low": 10.268608414239482,
      "ci_high": 18.36150793650794,
      "ci_metric": "octane_score",
      "median_latency_ms": 1430.6291248649359,
      "p95_latency_ms": 1463.2702121511102,
      "input_tokens": 360,
      "output_tokens": 558,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "454ff0af-07bf-4723-962f-2fa4d4413d0d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0035910000000000004,
      "cost_per_success_usd": 0.0011970000000000001,
      "octane_score": 15.382066276803119,
      "ci_low": 12.811131535947712,
      "ci_high": 18.815724815724817,
      "ci_metric": "octane_score",
      "median_latency_ms": 1681.1710419133306,
      "p95_latency_ms": 1750.288491230458,
      "input_tokens": 402,
      "output_tokens": 731,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "133d8098-9d2c-4058-aa88-f863897a92ec",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001692,
      "cost_per_success_usd": 0.0005639999999999999,
      "octane_score": 11.421985815602838,
      "ci_low": 9.734042553191491,
      "ci_high": 14.044326241134755,
      "ci_metric": "octane_score",
      "median_latency_ms": 1041.8688748031855,
      "p95_latency_ms": 1177.809413522482,
      "input_tokens": 258,
      "output_tokens": 333,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "82d80bd6-2f6a-4df9-80db-7cdec379d7ec",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0004005,
      "cost_per_success_usd": 0.0001335,
      "octane_score": 27.752808988764052,
      "ci_low": 24.56928838951311,
      "ci_high": 33.992509363295895,
      "ci_metric": "octane_score",
      "median_latency_ms": 742.2217493876815,
      "p95_latency_ms": 806.3501248136163,
      "input_tokens": 246,
      "output_tokens": 48,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "14182c3a-ba56-409e-a103-c4676e21bd82",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00201825,
      "cost_per_success_usd": 0.0006727500000000001,
      "octane_score": 30.397621701969523,
      "ci_low": 15.244890375325152,
      "ci_high": 70.47342995169082,
      "ci_metric": "octane_score",
      "median_latency_ms": 1318.8229575753212,
      "p95_latency_ms": 1595.10717112571,
      "input_tokens": 315,
      "output_tokens": 396,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8b44f3df-383b-4aab-9240-6ca204ed9ca5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00071775,
      "cost_per_success_usd": 0.00023925,
      "octane_score": 11.869035179380008,
      "ci_low": 7.16720604099245,
      "ci_high": 14.845019777058614,
      "ci_metric": "octane_score",
      "median_latency_ms": 812.8379574045539,
      "p95_latency_ms": 952.8152206912637,
      "input_tokens": 135,
      "output_tokens": 137,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f2714278-8558-4c8b-9c50-4a2de05d372f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0030996,
      "cost_per_success_usd": 0.0010332,
      "octane_score": 17.82068654019874,
      "ci_low": 13.791278333908641,
      "ci_high": 22.96532980755105,
      "ci_metric": "octane_score",
      "median_latency_ms": 8034.17241666466,
      "p95_latency_ms": 10024.025430344043,
      "input_tokens": 402,
      "output_tokens": 680,
      "reasoning_tokens": 1836,
      "undefined_reason": null
    },
    {
      "snapshot_id": "50a8520a-ff04-4526-9f03-46c264744b6b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0030468,
      "cost_per_success_usd": 0.0010156,
      "octane_score": 11.268215833005122,
      "ci_low": 7.856583762537279,
      "ci_high": 15.228432873915263,
      "ci_metric": "octane_score",
      "median_latency_ms": 7576.969875022769,
      "p95_latency_ms": 12541.242600698022,
      "input_tokens": 390,
      "output_tokens": 447,
      "reasoning_tokens": 2027,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f9f9f4b9-8f5f-46f7-ac29-f27bd1d2480d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0006786,
      "cost_per_success_usd": 0.0002262,
      "octane_score": 90.40671971706455,
      "ci_low": 46.15661566156614,
      "ci_high": 213.37083708370838,
      "ci_metric": "octane_score",
      "median_latency_ms": 2031.985375098884,
      "p95_latency_ms": 3249.433937389403,
      "input_tokens": 315,
      "output_tokens": 237,
      "reasoning_tokens": 276,
      "undefined_reason": null
    },
    {
      "snapshot_id": "31ae4d57-9016-47a2-a668-12e8ba2abe97",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005543999999999999,
      "cost_per_success_usd": 0.00018479999999999997,
      "octane_score": 34.859307359307365,
      "ci_low": 28.79175187457397,
      "ci_high": 42.46096799375489,
      "ci_metric": "octane_score",
      "median_latency_ms": 2375.6352495402098,
      "p95_latency_ms": 2906.8245873786504,
      "input_tokens": 258,
      "output_tokens": 301,
      "reasoning_tokens": 118,
      "undefined_reason": null
    },
    {
      "snapshot_id": "40162b4a-9288-4eed-a9d7-c299b9e9f89a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001068,
      "cost_per_success_usd": 3.56e-05,
      "octane_score": 104.07303370786519,
      "ci_low": 92.13483146067416,
      "ci_high": 127.4719101123596,
      "ci_metric": "octane_score",
      "median_latency_ms": 698.3496667817235,
      "p95_latency_ms": 762.3633283190429,
      "input_tokens": 246,
      "output_tokens": 48,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f4b500de-28b5-4fd1-88e4-73b1b17e43ec",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.000405,
      "cost_per_success_usd": 0.000135,
      "octane_score": 21.034567901234567,
      "ci_low": 12.564296520423605,
      "ci_high": 27.089238845144358,
      "ci_metric": "octane_score",
      "median_latency_ms": 2733.9183753356338,
      "p95_latency_ms": 2846.239650528878,
      "input_tokens": 135,
      "output_tokens": 118,
      "reasoning_tokens": 197,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7b392c98-141c-480e-a8d0-b741a10bdee6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0009923999999999998,
      "cost_per_success_usd": 0.00033079999999999996,
      "octane_score": 39.08806932688432,
      "ci_low": 28.845454545454544,
      "ci_high": 50.78584553657508,
      "ci_metric": "octane_score",
      "median_latency_ms": 3222.680250182748,
      "p95_latency_ms": 3407.069099508226,
      "input_tokens": 360,
      "output_tokens": 498,
      "reasoning_tokens": 269,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5833f043-3a09-45af-a656-b40bec7d10ab",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003549,
      "cost_per_success_usd": 0.001183,
      "octane_score": 17.2865595942519,
      "ci_low": 8.766811032596305,
      "ci_high": 40.52684066560292,
      "ci_metric": "octane_score",
      "median_latency_ms": 5082.224166952074,
      "p95_latency_ms": 5447.827904578298,
      "input_tokens": 7149,
      "output_tokens": 760,
      "reasoning_tokens": 1006,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fd27ff0f-1b7d-48d7-a627-44cf548366a7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0050852,
      "cost_per_success_usd": 0.0016950666666666666,
      "octane_score": 7.628215212774326,
      "ci_low": 5.716888212684176,
      "ci_high": 9.854111918379967,
      "ci_metric": "octane_score",
      "median_latency_ms": 6450.092041864991,
      "p95_latency_ms": 7504.173503443599,
      "input_tokens": 8284,
      "output_tokens": 1561,
      "reasoning_tokens": 1296,
      "undefined_reason": null
    },
    {
      "snapshot_id": "571aacba-3d5f-4bf7-b9ce-a55e30de5705",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0115318,
      "cost_per_success_usd": 0.0038439333333333335,
      "octane_score": 4.789972077212578,
      "ci_low": 4.050761249767654,
      "ci_high": 5.9657231887821345,
      "ci_metric": "octane_score",
      "median_latency_ms": 14747.427500784397,
      "p95_latency_ms": 16325.490799732506,
      "input_tokens": 12005,
      "output_tokens": 1795,
      "reasoning_tokens": 5814,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f2642de9-abbd-403d-8d08-492a32b70e46",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003085,
      "cost_per_success_usd": 0.0010283333333333334,
      "octane_score": 2.7614262560777956,
      "ci_low": 1.7357531001811346,
      "ci_high": 3.672169643492493,
      "ci_metric": "octane_score",
      "median_latency_ms": 4402.944874949753,
      "p95_latency_ms": 5242.164350673555,
      "input_tokens": 6179,
      "output_tokens": 405,
      "reasoning_tokens": 1136,
      "undefined_reason": null
    },
    {
      "snapshot_id": "45912e20-1ec2-4c93-9ba4-1591764ce590",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0034197999999999997,
      "cost_per_success_usd": 0.0011399333333333332,
      "octane_score": 5.651207672963332,
      "ci_low": 4.6420518602029315,
      "ci_high": 6.948651967951344,
      "ci_metric": "octane_score",
      "median_latency_ms": 5565.047457814217,
      "p95_latency_ms": 6604.656508285552,
      "input_tokens": 6605,
      "output_tokens": 913,
      "reasoning_tokens": 836,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8fa103fb-e14f-4243-8b74-cd94888c5518",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00092424,
      "cost_per_success_usd": 0.00030808,
      "octane_score": 12.026097117631785,
      "ci_low": 8.383600858807892,
      "ci_high": 18.33689015353608,
      "ci_metric": "octane_score",
      "median_latency_ms": 1813.6453330516815,
      "p95_latency_ms": 2104.715045262128,
      "input_tokens": 5514,
      "output_tokens": 195,
      "reasoning_tokens": 72,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ee2ebb19-be2f-4bfc-a73b-cbd104dcc782",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.013803,
      "cost_per_success_usd": 0.004601,
      "octane_score": 2.4872853727450557,
      "ci_low": 2.2387297175787793,
      "ci_high": 2.769577124820128,
      "ci_metric": "octane_score",
      "median_latency_ms": 22111.086499877274,
      "p95_latency_ms": 24955.661187600344,
      "input_tokens": 13923,
      "output_tokens": 1462,
      "reasoning_tokens": 7720,
      "undefined_reason": null
    },
    {
      "snapshot_id": "74558892-ab17-4b95-944e-8cde89fc9409",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00313,
      "cost_per_success_usd": 0.0010433333333333332,
      "octane_score": 2.721725239616614,
      "ci_low": 1.6154761904761905,
      "ci_high": 4.272139303482587,
      "ci_metric": "octane_score",
      "median_latency_ms": 3233.7110405787826,
      "p95_latency_ms": 4008.9209049008787,
      "input_tokens": 135,
      "output_tokens": 91,
      "reasoning_tokens": 195,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7ebf105c-aaed-4a8b-afe7-f58b10a4af83",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000972,
      "cost_per_success_usd": 0.000324,
      "octane_score": 11.435185185185187,
      "ci_low": 10.123456790123456,
      "ci_high": 14.006172839506174,
      "ci_metric": "octane_score",
      "median_latency_ms": 1134.0391663834453,
      "p95_latency_ms": 1215.1601791381836,
      "input_tokens": 246,
      "output_tokens": 48,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3705769b-93a4-40aa-89e6-3c7e79f822a7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003576,
      "cost_per_success_usd": 0.0011920000000000001,
      "octane_score": 5.404362416107382,
      "ci_low": 4.605704697986577,
      "ci_high": 6.64513422818792,
      "ci_metric": "octane_score",
      "median_latency_ms": 2006.0612922534347,
      "p95_latency_ms": 2277.3492662236094,
      "input_tokens": 258,
      "output_tokens": 306,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4041cfb9-34f8-43c4-83c8-940dab13a4ef",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.023264,
      "cost_per_success_usd": 0.007754666666666667,
      "octane_score": 2.3743552269601103,
      "ci_low": 1.9282496782496785,
      "ci_high": 2.92737003058104,
      "ci_metric": "octane_score",
      "median_latency_ms": 12550.15124939382,
      "p95_latency_ms": 15024.998100288212,
      "input_tokens": 402,
      "output_tokens": 662,
      "reasoning_tokens": 1584,
      "undefined_reason": null
    },
    {
      "snapshot_id": "905c5088-95a4-468e-be29-8e0ab631a957",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00723,
      "cost_per_success_usd": 0.0024100000000000002,
      "octane_score": 5.365283540802213,
      "ci_low": 3.8695121951219513,
      "ci_high": 7.344603174603177,
      "ci_metric": "octane_score",
      "median_latency_ms": 4319.761583581567,
      "p95_latency_ms": 4634.75423399359,
      "input_tokens": 360,
      "output_tokens": 518,
      "reasoning_tokens": 133,
      "undefined_reason": null
    },
    {
      "snapshot_id": "42f284ed-a59d-4c2f-8c36-e057f7b33b9b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.01738,
      "cost_per_success_usd": 0.0057933333333333335,
      "octane_score": 1.9753739930955123,
      "ci_low": 1.690524781341108,
      "ci_high": 2.2918693982074263,
      "ci_metric": "octane_score",
      "median_latency_ms": 10112.020209431648,
      "p95_latency_ms": 14027.009745966643,
      "input_tokens": 390,
      "output_tokens": 475,
      "reasoning_tokens": 1185,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3afec709-c5ce-4cb8-81b9-32a02689e092",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0035399999999999997,
      "cost_per_success_usd": 0.0011799999999999998,
      "octane_score": 17.330508474576273,
      "ci_low": 7.851369863013699,
      "ci_high": 52.678888888888885,
      "ci_metric": "octane_score",
      "median_latency_ms": 1726.4395831152797,
      "p95_latency_ms": 3992.2544205561276,
      "input_tokens": 315,
      "output_tokens": 226,
      "reasoning_tokens": 65,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c3195331-02ab-4d9b-a8aa-b608dfcffc6b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.04578,
      "cost_per_success_usd": 0.015260000000000003,
      "octane_score": 0.4221494102228046,
      "ci_low": 0.3174111126767179,
      "ci_high": 0.5675862518750406,
      "ci_metric": "octane_score",
      "median_latency_ms": 3234.579416923225,
      "p95_latency_ms": 3640.7426668331027,
      "input_tokens": 6357,
      "output_tokens": 963,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5fc939f4-0460-450b-b18b-4fe0c5f59efd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.020766,
      "cost_per_success_usd": 0.006922,
      "octane_score": 0.5352499277665416,
      "ci_low": 0.3644039551160982,
      "ci_high": 0.8516071305664534,
      "ci_metric": "octane_score",
      "median_latency_ms": 2885.4311667382717,
      "p95_latency_ms": 3010.25025350973,
      "input_tokens": 5514,
      "output_tokens": 189,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fb8a7bd6-c52c-4ea7-a296-85feef7bcc46",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.05364000000000001,
      "cost_per_success_usd": 0.017880000000000004,
      "octane_score": 1.1437360178970915,
      "ci_low": 0.5865599084929939,
      "ci_high": 2.7185206422018346,
      "ci_metric": "octane_score",
      "median_latency_ms": 3884.946250356734,
      "p95_latency_ms": 4900.992437452078,
      "input_tokens": 6306,
      "output_tokens": 687,
      "reasoning_tokens": 50,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cbec50f2-c775-4942-bc07-9712880b1ea7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.100065,
      "cost_per_success_usd": 0.033355,
      "octane_score": 0.3876580222855144,
      "ci_low": 0.2895547232625884,
      "ci_high": 0.508926700264458,
      "ci_metric": "octane_score",
      "median_latency_ms": 6995.770375244319,
      "p95_latency_ms": 7418.921875394881,
      "input_tokens": 7755,
      "output_tokens": 1532,
      "reasoning_tokens": 511,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d86f1d52-60ff-4ff2-9f25-6e8b89c9fff8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.399525,
      "cost_per_success_usd": 0.13317500000000002,
      "octane_score": 0.13825667980727113,
      "ci_low": 0.11274044114188964,
      "ci_high": 0.17233300411316838,
      "ci_metric": "octane_score",
      "median_latency_ms": 34853.68979163468,
      "p95_latency_ms": 41116.573192179196,
      "input_tokens": 14691,
      "output_tokens": 2164,
      "reasoning_tokens": 8705,
      "undefined_reason": null
    },
    {
      "snapshot_id": "faa6c0cc-99db-4571-b394-4b36069124ce",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.23921,
      "cost_per_success_usd": 0.07973666666666666,
      "octane_score": 0.14352242799214082,
      "ci_low": 0.12459580818177479,
      "ci_high": 0.16391108262173484,
      "ci_metric": "octane_score",
      "median_latency_ms": 20271.460874937475,
      "p95_latency_ms": 24923.08871289715,
      "input_tokens": 11362,
      "output_tokens": 1456,
      "reasoning_tokens": 4624,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2fd45a2b-23a3-47fd-a2c1-906a9e4f1145",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.068325,
      "cost_per_success_usd": 0.022775,
      "octane_score": 0.12468349798755946,
      "ci_low": 0.07548284480799818,
      "ci_high": 0.15720051785850278,
      "ci_metric": "octane_score",
      "median_latency_ms": 7771.50437515229,
      "p95_latency_ms": 8240.319813229144,
      "input_tokens": 5985,
      "output_tokens": 354,
      "reasoning_tokens": 926,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2c8a5e51-00d5-4467-8297-192e3395cd05",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001974,
      "cost_per_success_usd": 0.0006580000000000001,
      "octane_score": 4.315602836879432,
      "ci_low": 2.603448275862069,
      "ci_high": 5.3923719958202705,
      "ci_metric": "octane_score",
      "median_latency_ms": 1030.4155834019184,
      "p95_latency_ms": 1048.3562206849456,
      "input_tokens": 135,
      "output_tokens": 142,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9926c310-8249-46f2-8f16-8f58d51fffce",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.030636,
      "cost_per_success_usd": 0.010212,
      "octane_score": 1.803009531270401,
      "ci_low": 1.2110740628473275,
      "ci_high": 2.9319002123142255,
      "ci_metric": "octane_score",
      "median_latency_ms": 8379.440207965672,
      "p95_latency_ms": 14891.634969972072,
      "input_tokens": 402,
      "output_tokens": 558,
      "reasoning_tokens": 1928,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5dd6575f-5fc3-4f64-885d-26b8578f3854",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.006972000000000001,
      "cost_per_success_usd": 0.002324,
      "octane_score": 5.563826735513483,
      "ci_low": 4.124350086655111,
      "ci_high": 7.228392584597936,
      "ci_metric": "octane_score",
      "median_latency_ms": 1587.8317914903164,
      "p95_latency_ms": 1817.8382670506835,
      "input_tokens": 360,
      "output_tokens": 521,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5d4ec50f-32ee-4a3e-b6b0-1075bd79a3af",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001068,
      "cost_per_success_usd": 0.000356,
      "octane_score": 10.40730337078652,
      "ci_low": 9.213483146067416,
      "ci_high": 12.747191011235959,
      "ci_metric": "octane_score",
      "median_latency_ms": 666.6722502559423,
      "p95_latency_ms": 1071.4664872735739,
      "input_tokens": 246,
      "output_tokens": 48,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f188d1b2-547c-4f7b-8389-46aae01bb611",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004428,
      "cost_per_success_usd": 0.0014759999999999999,
      "octane_score": 4.36449864498645,
      "ci_low": 3.7195121951219523,
      "ci_high": 5.366531165311654,
      "ci_metric": "octane_score",
      "median_latency_ms": 1354.1844580322504,
      "p95_latency_ms": 1436.2155577167869,
      "input_tokens": 258,
      "output_tokens": 326,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e6966684-f631-4b0b-b067-0e8d0d25674a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.006437999999999999,
      "cost_per_success_usd": 0.002146,
      "octane_score": 9.529356943150047,
      "ci_low": 4.663547599674533,
      "ci_high": 24.979452054794525,
      "ci_metric": "octane_score",
      "median_latency_ms": 2845.303915441036,
      "p95_latency_ms": 3654.0456789545715,
      "input_tokens": 315,
      "output_tokens": 242,
      "reasoning_tokens": 242,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5e10a9dd-3c91-485d-b1f3-cb1b1e1637d1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.023639999999999998,
      "cost_per_success_usd": 0.00788,
      "octane_score": 1.452284263959391,
      "ci_low": 1.282186329588015,
      "ci_high": 1.7477092241905925,
      "ci_metric": "octane_score",
      "median_latency_ms": 8715.08566569537,
      "p95_latency_ms": 9415.20860325545,
      "input_tokens": 390,
      "output_tokens": 475,
      "reasoning_tokens": 1430,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cacbfb48-3f92-4b9b-97d2-88450e28a642",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0077491,
      "cost_per_success_usd": 0.0025830333333333334,
      "octane_score": 1.0993534733065775,
      "ci_low": 0.6599912583939287,
      "ci_high": 1.6826951794994327,
      "ci_metric": "octane_score",
      "median_latency_ms": 2663.7184573337436,
      "p95_latency_ms": 2869.785421155393,
      "input_tokens": 5371,
      "output_tokens": 487,
      "reasoning_tokens": 127,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0fce4123-bfa7-42e2-bc10-48fbf1b538b6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.024059,
      "cost_per_success_usd": 0.008019666666666666,
      "octane_score": 1.6123280269337879,
      "ci_low": 1.2019951174341272,
      "ci_high": 2.0934729064039406,
      "ci_metric": "octane_score",
      "median_latency_ms": 6591.833459213376,
      "p95_latency_ms": 6823.524621035904,
      "input_tokens": 7967,
      "output_tokens": 1508,
      "reasoning_tokens": 1174,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0cd6a439-6ae9-41ec-a78c-d185373b2042",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.016059,
      "cost_per_success_usd": 0.005353,
      "octane_score": 3.820287689146273,
      "ci_low": 1.9473417721518982,
      "ci_high": 9.010643015521064,
      "ci_metric": "octane_score",
      "median_latency_ms": 4690.696083940566,
      "p95_latency_ms": 5079.36202129349,
      "input_tokens": 6993,
      "output_tokens": 786,
      "reasoning_tokens": 725,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8a3e07c2-cbb3-47fb-8ca2-eb4b80420954",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.058573,
      "cost_per_success_usd": 0.019524333333333335,
      "octane_score": 0.5861403718436823,
      "ci_low": 0.5107464217446034,
      "ci_high": 0.6827347571888797,
      "ci_metric": "octane_score",
      "median_latency_ms": 17638.58770765364,
      "p95_latency_ms": 23717.395420186218,
      "input_tokens": 12391,
      "output_tokens": 1559,
      "reasoning_tokens": 6138,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d2cbc38d-a6d8-4d76-9ef3-96061f128ef4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.013226,
      "cost_per_success_usd": 0.004408666666666667,
      "octane_score": 1.4612127627400575,
      "ci_low": 1.247600059075469,
      "ci_high": 1.79127091813659,
      "ci_metric": "octane_score",
      "median_latency_ms": 3706.769000738859,
      "p95_latency_ms": 5397.499287314712,
      "input_tokens": 6362,
      "output_tokens": 974,
      "reasoning_tokens": 170,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8078af4d-798c-42d9-945f-797e260896d0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.062963,
      "cost_per_success_usd": 0.02098766666666667,
      "octane_score": 0.877293013357051,
      "ci_low": 0.7185528228646794,
      "ci_high": 1.0896414342629481,
      "ci_metric": "octane_score",
      "median_latency_ms": 20464.490374550223,
      "p95_latency_ms": 25262.670986633748,
      "input_tokens": 12917,
      "output_tokens": 1950,
      "reasoning_tokens": 6391,
      "undefined_reason": null
    },
    {
      "snapshot_id": "68e51e2b-840f-426a-b0c5-ec55e3eaa824",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0047231999999999994,
      "cost_per_success_usd": 0.0015743999999999999,
      "octane_score": 2.353277439024391,
      "ci_low": 1.6732986429956127,
      "ci_high": 3.469908575915644,
      "ci_metric": "octane_score",
      "median_latency_ms": 2181.3926259055734,
      "p95_latency_ms": 2221.6300124302506,
      "input_tokens": 5514,
      "output_tokens": 195,
      "reasoning_tokens": 89,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0ce07a9e-73ff-4eda-b7ad-20a2fccca202",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00055237,
      "cost_per_success_usd": 0.00018412333333333336,
      "octane_score": 100.0,
      "ci_low": 0.00015681,
      "ci_high": 0.00022974,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 9593.093291856349,
      "p95_latency_ms": 12460.874741151927,
      "input_tokens": 585,
      "output_tokens": 1077,
      "reasoning_tokens": 2069,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bc281a54-671a-45f0-8751-39082661e97e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 8.519e-05,
      "cost_per_success_usd": 2.8396666666666665e-05,
      "octane_score": 100.0,
      "ci_low": 1.6610000000000002e-05,
      "ci_high": 3.463e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 7864.909540861845,
      "p95_latency_ms": 7906.509904470295,
      "input_tokens": 318,
      "output_tokens": 164,
      "reasoning_tokens": 281,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6746c55e-5e87-4f9f-9f99-9b7ec694f42a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00038791000000000005,
      "cost_per_success_usd": 0.00012930333333333334,
      "octane_score": 100.0,
      "ci_low": 9.519e-05,
      "ci_high": 0.00016999,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 28023.45908433199,
      "p95_latency_ms": 38145.714158564806,
      "input_tokens": 543,
      "output_tokens": 801,
      "reasoning_tokens": 1385,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2d31c616-5ebc-4771-89d6-9f3d13bca86c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00011115000000000002,
      "cost_per_success_usd": 3.7050000000000006e-05,
      "octane_score": 100.0,
      "ci_low": 3.28e-05,
      "ci_high": 4.538000000000001e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 4027.7148755267262,
      "p95_latency_ms": 8765.321399550885,
      "input_tokens": 407,
      "output_tokens": 157,
      "reasoning_tokens": 425,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2d9536df-b9b4-4306-ac1f-4b6d72b3f384",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00019326,
      "cost_per_success_usd": 6.442e-05,
      "octane_score": 100.0,
      "ci_low": 5.4900000000000006e-05,
      "ci_high": 7.921000000000001e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 12991.170042194426,
      "p95_latency_ms": 17248.986228741705,
      "input_tokens": 441,
      "output_tokens": 625,
      "reasoning_tokens": 434,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1835e2a7-b8d7-4b0e-9b0f-d072977fff2a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00034332000000000004,
      "cost_per_success_usd": 0.00011444000000000001,
      "octane_score": 100.0,
      "ci_low": 0.00010802000000000001,
      "ci_high": 0.00012269,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 13898.526166565716,
      "p95_latency_ms": 30247.45895443484,
      "input_tokens": 547,
      "output_tokens": 666,
      "reasoning_tokens": 1257,
      "undefined_reason": null
    },
    {
      "snapshot_id": "30b5e475-7a63-4910-9e73-07b18bcc6f88",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.000409,
      "cost_per_success_usd": 0.0002045,
      "octane_score": 99.99999999999999,
      "ci_low": 0.00010255999999999998,
      "ci_high": 0.00047411,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 31715.533750131726,
      "p95_latency_ms": 38848.65861330181,
      "input_tokens": 498,
      "output_tokens": 1114,
      "reasoning_tokens": 1204,
      "undefined_reason": null
    },
    {
      "snapshot_id": "990b2311-3455-4215-ae64-fae5a64e743a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0046152,
      "cost_per_success_usd": 0.0015383999999999999,
      "octane_score": 2.408346333853355,
      "ci_low": 1.8964303549477568,
      "ci_high": 3.074801192842943,
      "ci_metric": "octane_score",
      "median_latency_ms": 2967.3480419442058,
      "p95_latency_ms": 3073.502329364419,
      "input_tokens": 864,
      "output_tokens": 36,
      "reasoning_tokens": 554,
      "undefined_reason": null
    },
    {
      "snapshot_id": "28cab960-4818-4369-b74a-6b9d7abd76d8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0206216,
      "cost_per_success_usd": 0.0068738666666666665,
      "octane_score": 2.9750358847034177,
      "ci_low": 1.4920277767001588,
      "ci_high": 7.501265742674514,
      "ci_metric": "octane_score",
      "median_latency_ms": 16047.867082990706,
      "p95_latency_ms": 21934.435470774766,
      "input_tokens": 933,
      "output_tokens": 419,
      "reasoning_tokens": 2852,
      "undefined_reason": null
    },
    {
      "snapshot_id": "066b293c-b59e-4108-b313-0f043f1ec7dd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0355736,
      "cost_per_success_usd": 0.011857866666666666,
      "octane_score": 1.5527526030539505,
      "ci_low": 1.1745913527497942,
      "ci_high": 2.0616823169445033,
      "ci_metric": "octane_score",
      "median_latency_ms": 27201.978833414614,
      "p95_latency_ms": 32720.6445094198,
      "input_tokens": 1014,
      "output_tokens": 957,
      "reasoning_tokens": 4779,
      "undefined_reason": null
    },
    {
      "snapshot_id": "187b902c-3366-4117-b4cd-55ef0d5a2022",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.016144,
      "cost_per_success_usd": 0.005381333333333333,
      "octane_score": 2.4028121902874138,
      "ci_low": 1.7944126074498563,
      "ci_high": 3.1145868959788925,
      "ci_metric": "octane_score",
      "median_latency_ms": 13193.480541929603,
      "p95_latency_ms": 13496.35007865727,
      "input_tokens": 972,
      "output_tokens": 487,
      "reasoning_tokens": 2061,
      "undefined_reason": null
    },
    {
      "snapshot_id": "806267c7-0e7b-45db-b35b-a4b40e58a940",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0038712000000000004,
      "cost_per_success_usd": 0.0012904000000000001,
      "octane_score": 2.200609630088861,
      "ci_low": 1.3929889298892988,
      "ci_high": 2.8852174885385216,
      "ci_metric": "octane_score",
      "median_latency_ms": 2440.528874285519,
      "p95_latency_ms": 3859.48638683185,
      "input_tokens": 738,
      "output_tokens": 124,
      "reasoning_tokens": 384,
      "undefined_reason": null
    },
    {
      "snapshot_id": "14d8d5be-7b08-4e28-a91e-94e97c2b0e79",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0063236,
      "cost_per_success_usd": 0.002107866666666667,
      "octane_score": 3.0561705357707627,
      "ci_low": 2.522696052081467,
      "ci_high": 3.7578278195964323,
      "ci_metric": "octane_score",
      "median_latency_ms": 4807.785041630268,
      "p95_latency_ms": 4907.737392093986,
      "input_tokens": 858,
      "output_tokens": 339,
      "reasoning_tokens": 574,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3c58edcb-59ef-4670-9257-4f1a06047690",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.023278,
      "cost_per_success_usd": 0.007759333333333334,
      "octane_score": 1.474868974997852,
      "ci_low": 1.195786639499345,
      "ci_high": 2.297346789289493,
      "ci_metric": "octane_score",
      "median_latency_ms": 17434.510000050068,
      "p95_latency_ms": 19070.51327507943,
      "input_tokens": 996,
      "output_tokens": 564,
      "reasoning_tokens": 3165,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c0dbabda-2bed-41b7-b9b1-6fd896eee3e1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.027066,
      "cost_per_success_usd": 0.009021999999999999,
      "octane_score": 1.2684548880514301,
      "ci_low": 1.1555907172995779,
      "ci_high": 1.3858823529411763,
      "ci_metric": "octane_score",
      "median_latency_ms": 19454.63841687888,
      "p95_latency_ms": 21612.761728372425,
      "input_tokens": 996,
      "output_tokens": 498,
      "reasoning_tokens": 3809,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fd601c0e-5e75-4cc8-8770-acc650254970",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.008586,
      "cost_per_success_usd": 0.002862,
      "octane_score": 2.2508735150244585,
      "ci_low": 1.864790286975718,
      "ci_high": 2.7676450034940605,
      "ci_metric": "octane_score",
      "median_latency_ms": 6531.106750480831,
      "p95_latency_ms": 6752.132200356573,
      "input_tokens": 858,
      "output_tokens": 323,
      "reasoning_tokens": 950,
      "undefined_reason": null
    },
    {
      "snapshot_id": "dbd9472a-9322-4078-8fc3-8bc0e280d5dd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00525,
      "cost_per_success_usd": 0.00175,
      "octane_score": 2.1171428571428574,
      "ci_low": 1.811282616685042,
      "ci_high": 2.593142857142858,
      "ci_metric": "octane_score",
      "median_latency_ms": 3930.933958850801,
      "p95_latency_ms": 4054.2870205827057,
      "input_tokens": 864,
      "output_tokens": 36,
      "reasoning_tokens": 711,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4ba469e1-f4bf-448c-87ea-e629eb2f30f8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.007044,
      "cost_per_success_usd": 0.002348,
      "octane_score": 1.2093980692788189,
      "ci_low": 0.7074105621805793,
      "ci_high": 1.8616522366522368,
      "ci_metric": "octane_score",
      "median_latency_ms": 4421.237290836871,
      "p95_latency_ms": 7411.180991586297,
      "input_tokens": 738,
      "output_tokens": 80,
      "reasoning_tokens": 944,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0d02c956-366c-42a5-96d1-e817d7083cbd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.02073,
      "cost_per_success_usd": 0.0069099999999999995,
      "octane_score": 1.8712493970091657,
      "ci_low": 1.2821867321867322,
      "ci_high": 2.7154342723004703,
      "ci_metric": "octane_score",
      "median_latency_ms": 13863.440833054483,
      "p95_latency_ms": 19104.02342090383,
      "input_tokens": 972,
      "output_tokens": 474,
      "reasoning_tokens": 2817,
      "undefined_reason": null
    },
    {
      "snapshot_id": "81028569-ca07-493b-b94a-d9f2db0c201f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.029165999999999997,
      "cost_per_success_usd": 0.009722,
      "octane_score": 2.1034766508948777,
      "ci_low": 1.0549269697593087,
      "ci_high": 5.397427140255008,
      "ci_metric": "octane_score",
      "median_latency_ms": 26953.020790591836,
      "p95_latency_ms": 27811.924279108644,
      "input_tokens": 933,
      "output_tokens": 362,
      "reasoning_tokens": 4316,
      "undefined_reason": null
    },
    {
      "snapshot_id": "60aee30c-20b6-4211-8476-ef59510c667f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.058164,
      "cost_per_success_usd": 0.019388,
      "octane_score": 0.9496767760126541,
      "ci_low": 0.6752295645315758,
      "ci_high": 1.3796222751743006,
      "ci_metric": "octane_score",
      "median_latency_ms": 37173.653666861355,
      "p95_latency_ms": 81956.34559188037,
      "input_tokens": 1014,
      "output_tokens": 656,
      "reasoning_tokens": 8860,
      "undefined_reason": null
    },
    {
      "snapshot_id": "23b13fab-f4b4-4e30-af0a-a142c6823e68",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.02470275,
      "cost_per_success_usd": 0.00823425,
      "octane_score": 1.5703109977634069,
      "ci_low": 1.1694903859082966,
      "ci_high": 2.0245581439664617,
      "ci_metric": "octane_score",
      "median_latency_ms": 98537.30491641909,
      "p95_latency_ms": 279371.40586646274,
      "input_tokens": 365,
      "output_tokens": 543,
      "reasoning_tokens": 10363,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cd5bfb4b-d40f-4acd-9a63-df3669ae4212",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0078030700000000005,
      "cost_per_success_usd": 0.002601023333333333,
      "octane_score": 1.0917497856612846,
      "ci_low": 0.6815971505053504,
      "ci_high": 1.4782826655208223,
      "ci_metric": "octane_score",
      "median_latency_ms": 19978.24991773814,
      "p95_latency_ms": 21851.18951657787,
      "input_tokens": 139,
      "output_tokens": 111,
      "reasoning_tokens": 3337,
      "undefined_reason": null
    },
    {
      "snapshot_id": "68ebbc89-a644-4c87-bbd8-7407415ac296",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0733779,
      "cost_per_success_usd": 0.0244593,
      "octane_score": 0.7527743366872043,
      "ci_low": 0.5819371769055518,
      "ci_high": 0.971431856067138,
      "ci_metric": "octane_score",
      "median_latency_ms": 151036.843707785,
      "p95_latency_ms": 215389.94608316568,
      "input_tokens": 407,
      "output_tokens": 1962,
      "reasoning_tokens": 30569,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fdb60d1c-5d6d-4687-a5a3-f405b70d597d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.026592750000000002,
      "cost_per_success_usd": 0.00886425,
      "octane_score": 1.2910285698169615,
      "ci_low": 1.067771386451321,
      "ci_high": 1.7065569125695288,
      "ci_metric": "octane_score",
      "median_latency_ms": 68395.43487504125,
      "p95_latency_ms": 78623.02227541804,
      "input_tokens": 395,
      "output_tokens": 548,
      "reasoning_tokens": 11192,
      "undefined_reason": null
    },
    {
      "snapshot_id": "25e38ea2-9e42-4480-8b5e-fc1dcc4341f9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.02411563,
      "cost_per_success_usd": 0.008038543333333334,
      "octane_score": 2.54399325250885,
      "ci_low": 1.2254523775500858,
      "ci_high": 6.58967523496332,
      "ci_metric": "octane_score",
      "median_latency_ms": 56109.52954273671,
      "p95_latency_ms": 82952.96790366991,
      "input_tokens": 321,
      "output_tokens": 458,
      "reasoning_tokens": 10232,
      "undefined_reason": null
    },
    {
      "snapshot_id": "920f5b04-b7c4-4cf5-8197-f812e4b46557",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.007594200000000001,
      "cost_per_success_usd": 0.0025314,
      "octane_score": 2.5448368491743696,
      "ci_low": 1.9366782443229367,
      "ci_high": 3.4839512181931256,
      "ci_metric": "octane_score",
      "median_latency_ms": 18860.430791974068,
      "p95_latency_ms": 24490.83971623331,
      "input_tokens": 266,
      "output_tokens": 377,
      "reasoning_tokens": 2945,
      "undefined_reason": null
    },
    {
      "snapshot_id": "86924df5-1adc-48da-b466-4eced248fa90",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00266445,
      "cost_per_success_usd": 0.00088815,
      "octane_score": 4.171592636378991,
      "ci_low": 3.1051907412649,
      "ci_high": 5.802208127789523,
      "ci_metric": "octane_score",
      "median_latency_ms": 7697.9860831052065,
      "p95_latency_ms": 8477.21069632098,
      "input_tokens": 251,
      "output_tokens": 218,
      "reasoning_tokens": 916,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4ba77238-4ca0-41c5-83a1-d3ed7a52d150",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.008865000000000001,
      "cost_per_success_usd": 0.0029550000000000006,
      "octane_score": 1.2538071065989849,
      "ci_low": 1.0852380323248338,
      "ci_high": 1.5357021996615907,
      "ci_metric": "octane_score",
      "median_latency_ms": 1419.054207392037,
      "p95_latency_ms": 1513.7724197469652,
      "input_tokens": 483,
      "output_tokens": 282,
      "reasoning_tokens": 270,
      "undefined_reason": null
    },
    {
      "snapshot_id": "896b7ed9-b7df-4ce5-8b89-4a25f1f63a01",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.01695,
      "cost_per_success_usd": 0.00565,
      "octane_score": 1.1401769911504425,
      "ci_low": 0.7721998465669352,
      "ci_high": 1.7862889983579637,
      "ci_metric": "octane_score",
      "median_latency_ms": 2886.9619574397802,
      "p95_latency_ms": 4454.9595839343965,
      "input_tokens": 498,
      "output_tokens": 243,
      "reasoning_tokens": 845,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3f2ac97f-a0dd-46ba-8d2e-1a03f1660921",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.014031000000000002,
      "cost_per_success_usd": 0.004677000000000001,
      "octane_score": 4.372460979260209,
      "ci_low": 2.226177555893205,
      "ci_high": 10.703707048358694,
      "ci_metric": "octane_score",
      "median_latency_ms": 2912.1207911521196,
      "p95_latency_ms": 3175.6100798025727,
      "input_tokens": 555,
      "output_tokens": 357,
      "reasoning_tokens": 525,
      "undefined_reason": null
    },
    {
      "snapshot_id": "19289625-2045-449e-ab8d-42a2e0dbb287",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0697164,
      "cost_per_success_usd": 0.023238799999999997,
      "octane_score": 0.4924522780866483,
      "ci_low": 0.36764514676408677,
      "ci_high": 0.6491723676298742,
      "ci_metric": "octane_score",
      "median_latency_ms": 9573.539415374398,
      "p95_latency_ms": 15785.431104805319,
      "input_tokens": 627,
      "output_tokens": 612,
      "reasoning_tokens": 3991,
      "undefined_reason": null
    },
    {
      "snapshot_id": "746613f2-ef14-4e18-9af8-8f40bbfa2257",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0716274,
      "cost_per_success_usd": 0.0238758,
      "octane_score": 0.7711713673817563,
      "ci_low": 0.5679101843362734,
      "ci_high": 1.0628388829398467,
      "ci_metric": "octane_score",
      "median_latency_ms": 7913.350749760866,
      "p95_latency_ms": 10042.609450500457,
      "input_tokens": 639,
      "output_tokens": 552,
      "reasoning_tokens": 4176,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2694a76c-93a0-48f6-841e-d7d10503e8d6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.015312599999999999,
      "cost_per_success_usd": 0.0051042,
      "octane_score": 0.5563392239071092,
      "ci_low": 0.36114976517655245,
      "ci_high": 0.7545914487921858,
      "ci_metric": "octane_score",
      "median_latency_ms": 2516.9219579547644,
      "p95_latency_ms": 3372.71647118032,
      "input_tokens": 372,
      "output_tokens": 11,
      "reasoning_tokens": 970,
      "undefined_reason": null
    },
    {
      "snapshot_id": "116e9068-42e7-48bc-b545-b4087266137c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0661764,
      "cost_per_success_usd": 0.0220588,
      "octane_score": 0.5861757363652299,
      "ci_low": 0.39784552752555863,
      "ci_high": 0.8624378860570275,
      "ci_metric": "octane_score",
      "median_latency_ms": 14940.287540666759,
      "p95_latency_ms": 20729.11199070513,
      "input_tokens": 597,
      "output_tokens": 748,
      "reasoning_tokens": 3625,
      "undefined_reason": null
    },
    {
      "snapshot_id": "60f62864-932b-4239-b73f-72bf219d63c9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00011100000000000001,
      "cost_per_success_usd": 3.7000000000000005e-05,
      "octane_score": 76.74774774774774,
      "ci_low": 44.89189189189189,
      "ci_high": 93.59459459459458,
      "ci_metric": "octane_score",
      "median_latency_ms": 578.8342086598277,
      "p95_latency_ms": 1509.6397086046634,
      "input_tokens": 147,
      "output_tokens": 102,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "822900d7-e84f-44ae-ba89-0bfe4069df56",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005040000000000001,
      "cost_per_success_usd": 0.00016800000000000004,
      "octane_score": 76.96626984126982,
      "ci_low": 56.66071428571427,
      "ci_high": 101.18452380952378,
      "ci_metric": "octane_score",
      "median_latency_ms": 7052.651165984571,
      "p95_latency_ms": 7155.820229370147,
      "input_tokens": 372,
      "output_tokens": 537,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2bba59ea-769f-44a6-bd22-bf8647a9f811",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 8.280000000000001e-05,
      "cost_per_success_usd": 2.7600000000000003e-05,
      "octane_score": 134.23913043478262,
      "ci_low": 118.84057971014491,
      "ci_high": 164.4202898550725,
      "ci_metric": "octane_score",
      "median_latency_ms": 275.70362482219934,
      "p95_latency_ms": 306.2867872416973,
      "input_tokens": 258,
      "output_tokens": 39,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d2a75321-5ee4-458c-873c-71a66bc33086",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00026760000000000005,
      "cost_per_success_usd": 8.920000000000003e-05,
      "octane_score": 72.21973094170401,
      "ci_low": 61.54708520179371,
      "ci_high": 88.80044843049326,
      "ci_metric": "octane_score",
      "median_latency_ms": 4306.9772915914655,
      "p95_latency_ms": 6765.491316840052,
      "input_tokens": 270,
      "output_tokens": 267,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9c07fc22-32e1-402a-9bad-a87c2a5d190e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0003942,
      "cost_per_success_usd": 0.0001314,
      "octane_score": 155.63165905631658,
      "ci_low": 78.05175038051749,
      "ci_high": 360.8143074581431,
      "ci_metric": "octane_score",
      "median_latency_ms": 5691.346208564937,
      "p95_latency_ms": 5966.194808296859,
      "input_tokens": 327,
      "output_tokens": 411,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "19c95233-6afb-4293-bbea-ba3d29b6def1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.0005996,
      "cost_per_success_usd": 0.0005996,
      "octane_score": 30.70769401823438,
      "ci_low": 26.267806267806264,
      "ci_high": 96.09777313848296,
      "ci_metric": "octane_score",
      "median_latency_ms": 2863.1545417010784,
      "p95_latency_ms": 7506.192517280578,
      "input_tokens": 414,
      "output_tokens": 646,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2b361d6a-4109-469d-b665-6a12dc82d486",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00037640000000000004,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 2022.2448334097862,
      "p95_latency_ms": 4820.831395499407,
      "input_tokens": 402,
      "output_tokens": 370,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "d82d7af0-e93f-4a2c-b9d8-b78f2338ddaf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.351375229656696,
      "p95_latency_ms": 31.05536289513111,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "c6360f19-e60e-4715-bf2c-fa890dd025aa",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.647916719317436,
      "p95_latency_ms": 30.841379426419735,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "b486aad8-9389-47fb-8dea-2c6f67956daf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 29.39483243972063,
      "p95_latency_ms": 35.322232730686665,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "a638eb7c-1c63-4553-a856-8beaf0549142",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 37.86133322864771,
      "p95_latency_ms": 39.34652013704181,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "1a909ba2-4d4b-4211-8ef5-a0c837b0b4e7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 28.915000148117542,
      "p95_latency_ms": 30.12021305039525,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "f5e70e9c-bde0-4a4f-aa2b-9a9e909c98c8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.388332903385162,
      "p95_latency_ms": 36.48575842380524,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "d4b6dc5b-1e02-4aa5-89ab-b76cf93af511",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 29.07854225486517,
      "p95_latency_ms": 34.15742991492152,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "6b707081-b315-4f5f-89f8-662c39a2df4b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.06522,
      "cost_per_success_usd": 0.02174,
      "octane_score": 0.5947715424716346,
      "ci_low": 0.42588411316648533,
      "ci_high": 0.8129128601546031,
      "ci_metric": "octane_score",
      "median_latency_ms": 3490.6638748943806,
      "p95_latency_ms": 4202.943638153374,
      "input_tokens": 507,
      "output_tokens": 892,
      "reasoning_tokens": 311,
      "undefined_reason": null
    },
    {
      "snapshot_id": "12bf481b-26ef-4d19-bcd0-1570ebda0d43",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.12203,
      "cost_per_success_usd": 0.040676666666666667,
      "octane_score": 0.45265098746209953,
      "ci_low": 0.35590104403086703,
      "ci_high": 0.5719381688466112,
      "ci_metric": "octane_score",
      "median_latency_ms": 4727.332540787756,
      "p95_latency_ms": 5327.297140937298,
      "input_tokens": 588,
      "output_tokens": 1229,
      "reasoning_tokens": 1094,
      "undefined_reason": null
    },
    {
      "snapshot_id": "79e39973-91fd-4424-931a-0cb54f6e9e64",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.14126,
      "cost_per_success_usd": 0.047086666666666666,
      "octane_score": 0.24304120062296478,
      "ci_low": 0.22481581043409002,
      "ci_high": 0.2634991903430002,
      "ci_metric": "octane_score",
      "median_latency_ms": 5679.937917739153,
      "p95_latency_ms": 5719.2391546443105,
      "input_tokens": 516,
      "output_tokens": 1177,
      "reasoning_tokens": 1545,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f3c15d42-c408-4f55-b25b-7b6b63bf7074",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.06047,
      "cost_per_success_usd": 0.030235,
      "octane_score": 0.6763684471638829,
      "ci_low": 0.27182831396727536,
      "ci_high": 1.8056747492700267,
      "ci_metric": "octane_score",
      "median_latency_ms": 4549.41666778177,
      "p95_latency_ms": 5054.426728654653,
      "input_tokens": 432,
      "output_tokens": 621,
      "reasoning_tokens": 502,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3644adf1-1e08-42ff-9435-308991829d1b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.02799,
      "cost_per_success_usd": 0.00933,
      "octane_score": 0.6904608788853162,
      "ci_low": 0.5884244372990355,
      "ci_high": 0.8489817792068597,
      "ci_metric": "octane_score",
      "median_latency_ms": 1725.7037088274956,
      "p95_latency_ms": 2106.6482084803283,
      "input_tokens": 369,
      "output_tokens": 486,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4f05704a-0bf0-458e-a66f-36a05401398f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.01589,
      "cost_per_success_usd": 0.005296666666666667,
      "octane_score": 0.5361233480176211,
      "ci_low": 0.3155794806839773,
      "ci_high": 0.653807426054122,
      "ci_metric": "octane_score",
      "median_latency_ms": 1447.9823326691985,
      "p95_latency_ms": 1934.6826457418501,
      "input_tokens": 204,
      "output_tokens": 277,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "60cc045b-4efe-4c3b-9a8d-3fae59dedc33",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00583,
      "cost_per_success_usd": 0.0019433333333333334,
      "octane_score": 1.9065180102915955,
      "ci_low": 1.5449843260188088,
      "ci_high": 2.3433712121212125,
      "ci_metric": "octane_score",
      "median_latency_ms": 1310.7507079839706,
      "p95_latency_ms": 2189.7476702928543,
      "input_tokens": 318,
      "output_tokens": 42,
      "reasoning_tokens": 11,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6f94e3bb-770f-4e85-8434-02164213f29c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0051176450000000005,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 34551.32016725838,
      "p95_latency_ms": 36327.57869139314,
      "input_tokens": 393,
      "output_tokens": 19585,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "d1736622-5b33-4921-b99b-388184ff9486",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000569075,
      "cost_per_success_usd": 0.0001896916666666667,
      "octane_score": 19.531696173615078,
      "ci_low": 17.05614147409221,
      "ci_high": 23.923032992136363,
      "ci_metric": "octane_score",
      "median_latency_ms": 3572.0689995214343,
      "p95_latency_ms": 3913.771062158048,
      "input_tokens": 279,
      "output_tokens": 2119,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bd0c5e1c-5313-4275-86eb-2ace89333cc9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0020657,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 17292.95925050974,
      "p95_latency_ms": 18430.18912533298,
      "input_tokens": 276,
      "output_tokens": 7876,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "effcd77b-46c6-4a45-b71e-300dcc028df6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.004098185000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30353.744208812714,
      "p95_latency_ms": 30749.570096097887,
      "input_tokens": 345,
      "output_tokens": 15676,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "50aff9be-a6ce-4961-a558-d0857049abc0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.005673330000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 40927.12733335793,
      "p95_latency_ms": 45585.95993341878,
      "input_tokens": 426,
      "output_tokens": 21714,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "78b2ded0-997d-437e-9965-1410c596e545",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0008698950000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 7526.148917153478,
      "p95_latency_ms": 7593.8895164988935,
      "input_tokens": 423,
      "output_tokens": 3240,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "c3432133-51dd-4ed1-8b4d-183502ba26e9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0025762100000000002,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 19462.106375023723,
      "p95_latency_ms": 22784.78187425062,
      "input_tokens": 150,
      "output_tokens": 9871,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "4fd9f91f-8317-4ade-a375-5b8d5b58f8e4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.298128,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 130224.6026257053,
      "p95_latency_ms": 162279.3405005708,
      "input_tokens": 549,
      "output_tokens": 232,
      "reasoning_tokens": 49273,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "bcb6b8f5-f1fa-4da4-b922-05d42c8213b7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.020682,
      "cost_per_success_usd": 0.006894,
      "octane_score": 2.9663475485929793,
      "ci_low": 1.1513660104459624,
      "ci_high": 11.427453027139878,
      "ci_metric": "octane_score",
      "median_latency_ms": 18029.379457235336,
      "p95_latency_ms": 21323.212283663448,
      "input_tokens": 480,
      "output_tokens": 252,
      "reasoning_tokens": 3035,
      "undefined_reason": null
    },
    {
      "snapshot_id": "99cfca7b-8a9d-4f16-aae0-80ce08a99229",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.033672,
      "cost_per_success_usd": 0.011224,
      "octane_score": 1.1520254217153718,
      "ci_low": 0.7586131705189707,
      "ci_high": 2.204284577792931,
      "ci_metric": "octane_score",
      "median_latency_ms": 26616.228790953755,
      "p95_latency_ms": 46175.34315492957,
      "input_tokens": 537,
      "output_tokens": 943,
      "reasoning_tokens": 4490,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e4b716d5-09a8-4bb5-9ec2-b5e5387d24cb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0597072,
      "cost_per_success_usd": 0.0199024,
      "octane_score": 0.5750060294235871,
      "ci_low": 0.34556019070321814,
      "ci_high": 1.08843602992211,
      "ci_metric": "octane_score",
      "median_latency_ms": 21658.554000779986,
      "p95_latency_ms": 34842.94140050188,
      "input_tokens": 552,
      "output_tokens": 385,
      "reasoning_tokens": 9459,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e8956e90-a163-40b4-a17d-c1d92ee94c36",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.005412,
      "cost_per_success_usd": 0.001804,
      "octane_score": 1.5740946045824094,
      "ci_low": 0.9675885911840966,
      "ci_high": 2.225312634756361,
      "ci_metric": "octane_score",
      "median_latency_ms": 5471.25337459147,
      "p95_latency_ms": 8915.890312474221,
      "input_tokens": 294,
      "output_tokens": 131,
      "reasoning_tokens": 673,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c1b73468-d625-466e-a18e-38be35a14660",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0021516,
      "cost_per_success_usd": 0.0007172,
      "octane_score": 5.165923034021195,
      "ci_low": 4.109137902284475,
      "ci_high": 6.497059441293847,
      "ci_metric": "octane_score",
      "median_latency_ms": 2964.5282914862037,
      "p95_latency_ms": 4323.397191613913,
      "input_tokens": 405,
      "output_tokens": 45,
      "reasoning_tokens": 259,
      "undefined_reason": null
    },
    {
      "snapshot_id": "afeddf3a-026f-436e-963e-aaa38833bf59",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00924,
      "cost_per_success_usd": 0.0030800000000000003,
      "octane_score": 2.0915584415584414,
      "ci_low": 1.775430853299706,
      "ci_high": 2.571753246753247,
      "ci_metric": "octane_score",
      "median_latency_ms": 2915.936124511063,
      "p95_latency_ms": 4651.745612360537,
      "input_tokens": 402,
      "output_tokens": 190,
      "reasoning_tokens": 1216,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9d4d7fd7-23ec-4b47-b2cc-f538a185c5b9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.005673500000000001,
      "cost_per_success_usd": 0.0018911666666666669,
      "octane_score": 3.4063629153080104,
      "ci_low": 2.789104416013207,
      "ci_high": 4.188419846655504,
      "ci_metric": "octane_score",
      "median_latency_ms": 9111.548790708184,
      "p95_latency_ms": 10585.040341597049,
      "input_tokens": 398,
      "output_tokens": 703,
      "reasoning_tokens": 1014,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d9205850-152b-4fb8-86ab-759691df8bd9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00949935,
      "cost_per_success_usd": 0.00316645,
      "octane_score": 6.458336623032102,
      "ci_low": 2.675572533931899,
      "ci_high": 21.384079923624526,
      "ci_metric": "octane_score",
      "median_latency_ms": 12438.007917255163,
      "p95_latency_ms": 24492.366829700764,
      "input_tokens": 459,
      "output_tokens": 223,
      "reasoning_tokens": 2681,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ccd008e0-9798-47f6-9fab-bd657449786c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00088305,
      "cost_per_success_usd": 0.00029435,
      "octane_score": 12.587056225581792,
      "ci_low": 10.847961261211688,
      "ci_high": 15.035617648682978,
      "ci_metric": "octane_score",
      "median_latency_ms": 1944.5968326181173,
      "p95_latency_ms": 2072.7142334915698,
      "input_tokens": 405,
      "output_tokens": 106,
      "reasoning_tokens": 113,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6c8014af-4a82-417e-a2ab-951b62c83a8b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.12141460000000001,
      "cost_per_success_usd": 0.1214146,
      "octane_score": 0.09425555081514086,
      "ci_low": 0.09022802858964243,
      "ci_high": 0.6981709921056164,
      "ci_metric": "octane_score",
      "median_latency_ms": 210074.4151249528,
      "p95_latency_ms": 210251.32786314934,
      "input_tokens": 548,
      "output_tokens": 21694,
      "reasoning_tokens": 16171,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3b637c09-7111-4050-965a-374530b26c0b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.14660185,
      "cost_per_success_usd": 0.073300925,
      "octane_score": 0.2511882808209219,
      "ci_low": 0.09711327340324019,
      "ci_high": 0.5853258966203648,
      "ci_metric": "octane_score",
      "median_latency_ms": 210011.30533311516,
      "p95_latency_ms": 314900.3118212335,
      "input_tokens": 541,
      "output_tokens": 1350,
      "reasoning_tokens": 44387,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1b53b9cc-69cf-4d85-bf78-86c0b61f208b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004224800000000001,
      "cost_per_success_usd": 0.001408266666666667,
      "octane_score": 2.0164268131035783,
      "ci_low": 1.0283431109823518,
      "ci_high": 4.990812379110251,
      "ci_metric": "octane_score",
      "median_latency_ms": 4315.141000784934,
      "p95_latency_ms": 9002.238250058144,
      "input_tokens": 272,
      "output_tokens": 100,
      "reasoning_tokens": 1182,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cd6a85d0-04fb-492d-9445-ddc639197e4e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.025855550000000005,
      "cost_per_success_usd": 0.008618516666666668,
      "octane_score": 1.5002968414905116,
      "ci_low": 1.1007268221087507,
      "ci_high": 1.9835356853227226,
      "ci_metric": "octane_score",
      "median_latency_ms": 49662.384582683444,
      "p95_latency_ms": 69727.39187004045,
      "input_tokens": 511,
      "output_tokens": 1067,
      "reasoning_tokens": 6941,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2c506a2a-dd62-4f04-ba83-8c2f811b2336",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.296022,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 344686.63816712797,
      "p95_latency_ms": 348180.37389153615,
      "input_tokens": 537,
      "output_tokens": 6,
      "reasoning_tokens": 49152,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "97cfda9c-342f-47ae-ad72-7f2f7baf4907",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.065874,
      "cost_per_success_usd": 0.021958000000000002,
      "octane_score": 0.5211767920575644,
      "ci_low": 0.33364980723862137,
      "ci_high": 0.7744678089304258,
      "ci_metric": "octane_score",
      "median_latency_ms": 54562.66741733998,
      "p95_latency_ms": 108647.51892937346,
      "input_tokens": 540,
      "output_tokens": 634,
      "reasoning_tokens": 10165,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d25e8dbe-7f49-45cc-93e3-ee7b0fb15753",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.007914000000000001,
      "cost_per_success_usd": 0.002638,
      "octane_score": 2.4420015163002273,
      "ci_low": 1.9194501249715974,
      "ci_high": 3.0966410475377164,
      "ci_metric": "octane_score",
      "median_latency_ms": 7617.339584045112,
      "p95_latency_ms": 8232.928257901222,
      "input_tokens": 390,
      "output_tokens": 319,
      "reasoning_tokens": 870,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d33eb0b3-0b73-42a1-b4f9-5fe9774dff75",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0023940000000000003,
      "cost_per_success_usd": 0.0007980000000000001,
      "octane_score": 4.642857142857143,
      "ci_low": 3.883766745468873,
      "ci_high": 5.686716791979951,
      "ci_metric": "octane_score",
      "median_latency_ms": 2470.774500630796,
      "p95_latency_ms": 3375.5534259602427,
      "input_tokens": 393,
      "output_tokens": 51,
      "reasoning_tokens": 217,
      "undefined_reason": null
    },
    {
      "snapshot_id": "144e799f-a5b1-4eea-b823-61d40c03ed06",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.016361999999999998,
      "cost_per_success_usd": 0.005454,
      "octane_score": 3.7495416208287495,
      "ci_low": 1.7592081031307552,
      "ci_high": 10.066029723991509,
      "ci_metric": "octane_score",
      "median_latency_ms": 15579.938707873225,
      "p95_latency_ms": 21116.63352148607,
      "input_tokens": 459,
      "output_tokens": 290,
      "reasoning_tokens": 2284,
      "undefined_reason": null
    },
    {
      "snapshot_id": "84f0f650-90e5-4dbf-a711-ab25f4b0863d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00447,
      "cost_per_success_usd": 0.00149,
      "octane_score": 1.9058165548098434,
      "ci_low": 1.1361149110807116,
      "ci_high": 2.3891203703703705,
      "ci_metric": "octane_score",
      "median_latency_ms": 4250.795000232756,
      "p95_latency_ms": 4836.683975160122,
      "input_tokens": 264,
      "output_tokens": 161,
      "reasoning_tokens": 496,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e57f9113-cbc9-44bd-acc6-1a975dad1575",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.033714,
      "cost_per_success_usd": 0.011238,
      "octane_score": 1.1505902592394852,
      "ci_low": 0.6423559822747414,
      "ci_high": 3.0832477754962357,
      "ci_metric": "octane_score",
      "median_latency_ms": 33155.02495877445,
      "p95_latency_ms": 59609.956232830875,
      "input_tokens": 507,
      "output_tokens": 500,
      "reasoning_tokens": 4950,
      "undefined_reason": null
    },
    {
      "snapshot_id": "32c6a626-ac83-4d6a-b16c-c77f039df9ef",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0279825,
      "cost_per_success_usd": 0.0093275,
      "octane_score": 0.6906459394264272,
      "ci_low": 0.5156281266258455,
      "ci_high": 0.9369334452985381,
      "ci_metric": "octane_score",
      "median_latency_ms": 36416.65779147297,
      "p95_latency_ms": 58014.677716884755,
      "input_tokens": 351,
      "output_tokens": 510,
      "reasoning_tokens": 8759,
      "undefined_reason": null
    },
    {
      "snapshot_id": "69abf37b-420d-4a57-bd2d-5aabeb46b6ab",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.012884999999999999,
      "cost_per_success_usd": 0.004294999999999999,
      "octane_score": 0.6611563833915406,
      "ci_low": 0.38672875436554144,
      "ci_high": 1.0704210744658786,
      "ci_metric": "octane_score",
      "median_latency_ms": 20878.124499693513,
      "p95_latency_ms": 27367.146362178028,
      "input_tokens": 216,
      "output_tokens": 159,
      "reasoning_tokens": 4100,
      "undefined_reason": null
    },
    {
      "snapshot_id": "93c3ce4f-71e7-4cea-a5c4-55bebe519279",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.135375,
      "cost_per_success_usd": 0.0676875,
      "octane_score": 0.16907109879963067,
      "ci_low": 0.07769636253242812,
      "ci_high": 0.2884581453381393,
      "ci_metric": "octane_score",
      "median_latency_ms": 212430.50662521273,
      "p95_latency_ms": 227927.74833682925,
      "input_tokens": 492,
      "output_tokens": 368,
      "reasoning_tokens": 44675,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fd1d562b-aaf3-417d-8cdf-b4d3e350810e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0957075,
      "cost_per_success_usd": 0.0319025,
      "octane_score": 0.405307838988585,
      "ci_low": 0.29031639501438167,
      "ci_high": 0.566536269414192,
      "ci_metric": "octane_score",
      "median_latency_ms": 148732.7930405736,
      "p95_latency_ms": 176355.2317669615,
      "input_tokens": 459,
      "output_tokens": 636,
      "reasoning_tokens": 31190,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d4f31002-cbea-4943-9eca-f69fd93470d9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0028380000000000002,
      "cost_per_success_usd": 0.0009460000000000001,
      "octane_score": 3.9164904862579286,
      "ci_low": 3.096764059063776,
      "ci_high": 4.970427163198249,
      "ci_metric": "octane_score",
      "median_latency_ms": 4533.8955000042915,
      "p95_latency_ms": 5556.5070365555575,
      "input_tokens": 348,
      "output_tokens": 36,
      "reasoning_tokens": 852,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ee401667-83ae-4c7c-b265-ea6a5cf2ae71",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.120501,
      "cost_per_success_usd": 0.0602505,
      "octane_score": 0.339416270404395,
      "ci_low": 0.08571300388075104,
      "ci_high": 1.3259963641448749,
      "ci_metric": "octane_score",
      "median_latency_ms": 167408.83066691458,
      "p95_latency_ms": 224225.12645451352,
      "input_tokens": 414,
      "output_tokens": 481,
      "reasoning_tokens": 39617,
      "undefined_reason": null
    },
    {
      "snapshot_id": "376a6144-e267-47c3-90e1-adab5d9194a3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.14770650000000002,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 230517.80112460256,
      "p95_latency_ms": 239045.07196219638,
      "input_tokens": 501,
      "output_tokens": 0,
      "reasoning_tokens": 49152,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "4acd2e21-57eb-427c-b10e-fceba7edcba5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-implementation",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.003468,
      "cost_per_success_usd": 0.001156,
      "octane_score": 2.4564590542099194,
      "ci_low": 1.4368512110726646,
      "ci_high": 4.0347928068803744,
      "ci_metric": "octane_score",
      "median_latency_ms": 7517.134209163487,
      "p95_latency_ms": 11403.783420473335,
      "input_tokens": 261,
      "output_tokens": 144,
      "reasoning_tokens": 1191,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b7eb06eb-ea10-4c12-afea-73e24dff3ae5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-add-months",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.075172,
      "cost_per_success_usd": 0.025057333333333334,
      "octane_score": 0.5160299047517692,
      "ci_low": 0.3850494508828347,
      "ci_high": 0.6769408438364893,
      "ci_metric": "octane_score",
      "median_latency_ms": 140207.14274980128,
      "p95_latency_ms": 144308.89582438394,
      "input_tokens": 504,
      "output_tokens": 693,
      "reasoning_tokens": 29275,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b06ac174-2138-4ec2-952b-edcce7244232",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-parse-duration-contract",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.086446,
      "cost_per_success_usd": 0.028815333333333328,
      "octane_score": 0.39714966568724996,
      "ci_low": 0.3733768078072278,
      "ci_high": 0.42505499835437993,
      "ci_metric": "octane_score",
      "median_latency_ms": 158411.631250754,
      "p95_latency_ms": 177041.6850619018,
      "input_tokens": 537,
      "output_tokens": 551,
      "reasoning_tokens": 33920,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2461361b-d632-4c80-ba21-c04997278b29",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-dedup-preserve-order",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0019165,
      "cost_per_success_usd": 0.0006388333333333334,
      "octane_score": 5.7996347508479005,
      "ci_low": 4.961993455826831,
      "ci_high": 7.04840797307792,
      "ci_metric": "octane_score",
      "median_latency_ms": 5498.102916404605,
      "p95_latency_ms": 6973.056442197412,
      "input_tokens": 393,
      "output_tokens": 36,
      "reasoning_tokens": 652,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ffd0e988-619e-4f5d-bb58-3fa494230b63",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-subtract-intervals",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.12315300000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 213743.24529152364,
      "p95_latency_ms": 261730.7642409578,
      "input_tokens": 546,
      "output_tokens": 0,
      "reasoning_tokens": 49152,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "fd956d5c-1280-4997-8127-4ee2666aead9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-merge-intervals",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.026808,
      "cost_per_success_usd": 0.008936,
      "octane_score": 0.7209042076991943,
      "ci_low": 0.5668892393383218,
      "ci_high": 0.9136606055515893,
      "ci_metric": "octane_score",
      "median_latency_ms": 57450.391500256956,
      "p95_latency_ms": 59751.45236290991,
      "input_tokens": 396,
      "output_tokens": 365,
      "reasoning_tokens": 10279,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3ea4d4a0-0d52-4544-ad1b-e2bde807dc77",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "coding",
      "task_id": "coding-pair-sums-performance",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0769595,
      "cost_per_success_usd": 0.03847975,
      "octane_score": 0.5314483592019179,
      "ci_low": 0.21739130434782608,
      "ci_high": 1.383818335716996,
      "ci_metric": "octane_score",
      "median_latency_ms": 156851.28525085747,
      "p95_latency_ms": 205184.78808775544,
      "input_tokens": 459,
      "output_tokens": 343,
      "reasoning_tokens": 30349,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4a8dced6-9ea7-41a8-b03c-eeeb12810a84",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000375,
      "cost_per_success_usd": 0.000125,
      "octane_score": 9.509333333333332,
      "ci_low": 8.784,
      "ci_high": 10.280000000000001,
      "ci_metric": "octane_score",
      "median_latency_ms": 629.6560829505324,
      "p95_latency_ms": 849.8805338516831,
      "input_tokens": 285,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "749f0388-8e00-4adb-8bce-92d204629736",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.0014619999999999998,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 1022.6468751206994,
      "p95_latency_ms": 1147.2918504849076,
      "input_tokens": 537,
      "output_tokens": 185,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "a98456bf-53e2-4c8d-aa15-7090e7e4bc01",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.00209,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 1208.6208742111921,
      "p95_latency_ms": 1284.4557743519545,
      "input_tokens": 555,
      "output_tokens": 307,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "18938697-53dd-42a1-b5e3-9c8e7472d578",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.000567,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 720.9427915513515,
      "p95_latency_ms": 783.5018291138113,
      "input_tokens": 267,
      "output_tokens": 60,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "cfc30253-ef76-431b-9506-72eadb3747e4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0015999999999999999,
      "cost_per_success_usd": 0.0005333333333333333,
      "octane_score": 3.4650000000000003,
      "ci_low": 3.2641509433962264,
      "ci_high": 3.6600000000000006,
      "ci_metric": "octane_score",
      "median_latency_ms": 1211.95116546005,
      "p95_latency_ms": 1288.262842502445,
      "input_tokens": 315,
      "output_tokens": 257,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "857e8231-6008-4092-9576-a0d9536425e9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0029249999999999996,
      "cost_per_success_usd": 0.0009749999999999998,
      "octane_score": 1.8953846153846154,
      "ci_low": 1.7743589743589747,
      "ci_high": 2.002051282051282,
      "ci_metric": "octane_score",
      "median_latency_ms": 1693.4900414198637,
      "p95_latency_ms": 1735.6418420560658,
      "input_tokens": 315,
      "output_tokens": 54,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5dc5bc8c-1ece-4ecd-9b2a-b55d21e70aec",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001875,
      "cost_per_success_usd": 0.000625,
      "octane_score": 1.9018666666666666,
      "ci_low": 1.7568,
      "ci_high": 2.056,
      "ci_metric": "octane_score",
      "median_latency_ms": 1524.2665000259876,
      "p95_latency_ms": 1862.0528870262206,
      "input_tokens": 285,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a214ea05-8935-4d09-868c-67543c5475e8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0031349999999999998,
      "cost_per_success_usd": 0.001045,
      "octane_score": 2.392982456140351,
      "ci_low": 2.289952153110048,
      "ci_high": 2.566507177033493,
      "ci_metric": "octane_score",
      "median_latency_ms": 1749.2850832641125,
      "p95_latency_ms": 2027.1131704561412,
      "input_tokens": 537,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fa6ba630-ba49-4803-be8b-c7b982c98454",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.01796,
      "cost_per_success_usd": 0.005986666666666666,
      "octane_score": 2.518763919821826,
      "ci_low": 1.555592935239697,
      "ci_high": 3.8990825688073394,
      "ci_metric": "octane_score",
      "median_latency_ms": 4995.870833285153,
      "p95_latency_ms": 5358.282496593893,
      "input_tokens": 267,
      "output_tokens": 665,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d21c7d6e-4f95-485d-b26f-55bdb646017a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.01615,
      "cost_per_success_usd": 0.005383333333333334,
      "octane_score": 0.6297213622291021,
      "ci_low": 0.5175232198142415,
      "ci_high": 0.74314381270903,
      "ci_metric": "octane_score",
      "median_latency_ms": 4565.577667206526,
      "p95_latency_ms": 5295.760866347699,
      "input_tokens": 555,
      "output_tokens": 535,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "81664844-324a-48a1-b88a-c20d1636ff11",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00198,
      "cost_per_success_usd": 0.00066,
      "octane_score": 2.8,
      "ci_low": 2.621212121212121,
      "ci_high": 2.9575757575757575,
      "ci_metric": "octane_score",
      "median_latency_ms": 1240.4738329350948,
      "p95_latency_ms": 1368.9629334956408,
      "input_tokens": 315,
      "output_tokens": 69,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "472f9de5-e8b2-4603-9d7a-674144629b9e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0011250000000000001,
      "cost_per_success_usd": 0.00037500000000000006,
      "octane_score": 3.169777777777777,
      "ci_low": 2.9279999999999995,
      "ci_high": 3.4266666666666663,
      "ci_metric": "octane_score",
      "median_latency_ms": 1300.0797079876065,
      "p95_latency_ms": 1759.391107223928,
      "input_tokens": 285,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "596612c4-02e9-49a7-876a-12c56b8da1fd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.010206,
      "cost_per_success_usd": 0.003402,
      "octane_score": 4.432392710170488,
      "ci_low": 2.7304399173309712,
      "ci_high": 6.870958259847148,
      "ci_metric": "octane_score",
      "median_latency_ms": 4340.157083235681,
      "p95_latency_ms": 4406.226345337927,
      "input_tokens": 267,
      "output_tokens": 627,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f24a2603-bbf1-4937-86f9-7c3ac0919533",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.006135,
      "cost_per_success_usd": 0.002045,
      "octane_score": 1.6577017114914425,
      "ci_low": 1.3792079207920798,
      "ci_high": 1.8333333333333333,
      "ci_metric": "octane_score",
      "median_latency_ms": 3084.7110832110047,
      "p95_latency_ms": 3201.083596237004,
      "input_tokens": 555,
      "output_tokens": 298,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9db9d5ca-8096-4698-a3f0-17f7925c5090",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001971,
      "cost_per_success_usd": 0.000657,
      "octane_score": 3.8061897513952307,
      "ci_low": 3.6423135464231353,
      "ci_high": 4.082191780821917,
      "ci_metric": "octane_score",
      "median_latency_ms": 1369.0997092053294,
      "p95_latency_ms": 1577.004320640117,
      "input_tokens": 537,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4571f3cd-712a-4e4b-b3a9-71fbd89d9aba",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000129,
      "cost_per_success_usd": 4.2999999999999995e-05,
      "octane_score": 58.15503875968993,
      "ci_low": 55.65116279069768,
      "ci_high": 62.37209302325582,
      "ci_metric": "octane_score",
      "median_latency_ms": 956.0142504051328,
      "p95_latency_ms": 1200.8386257104576,
      "input_tokens": 483,
      "output_tokens": 9,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "55fde612-f6ad-45e7-aab7-5f7f1b01fdce",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 7.24e-05,
      "cost_per_success_usd": 2.413333333333333e-05,
      "octane_score": 624.8204419889503,
      "ci_low": 392.8030303030303,
      "ci_high": 968.5773480662984,
      "ci_metric": "octane_score",
      "median_latency_ms": 1018.3488754555583,
      "p95_latency_ms": 1043.1493871845305,
      "input_tokens": 252,
      "output_tokens": 40,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5ccc6040-506e-4096-8822-74e87b305bad",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00010188000000000002,
      "cost_per_success_usd": 3.396000000000001e-05,
      "octane_score": 54.41696113074203,
      "ci_low": 50.94228504122496,
      "ci_high": 57.47938751472319,
      "ci_metric": "octane_score",
      "median_latency_ms": 1638.400500640273,
      "p95_latency_ms": 1913.4472494013608,
      "input_tokens": 291,
      "output_tokens": 69,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "805cdc42-000b-459f-93df-bdfbacd646b1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 7.07e-05,
      "cost_per_success_usd": 2.3566666666666666e-05,
      "octane_score": 50.438472418670436,
      "ci_low": 42.76132624651987,
      "ci_high": 64.39141694413614,
      "ci_metric": "octane_score",
      "median_latency_ms": 1331.3580006361008,
      "p95_latency_ms": 1592.2383749857545,
      "input_tokens": 282,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6024f67c-64e1-4863-91aa-3bf0399e0b52",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00013674,
      "cost_per_success_usd": 4.558e-05,
      "octane_score": 74.37472575691092,
      "ci_low": 61.12329969284775,
      "ci_high": 81.7463799912242,
      "ci_metric": "octane_score",
      "median_latency_ms": 1323.3130415901542,
      "p95_latency_ms": 1449.134203325957,
      "input_tokens": 504,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e921856d-b211-4b49-8364-4e94a7e106d1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00014152,
      "cost_per_success_usd": 4.7173333333333324e-05,
      "octane_score": 319.65093273035615,
      "ci_low": 192.17939214232766,
      "ci_high": 499.89307100085546,
      "ci_metric": "octane_score",
      "median_latency_ms": 2487.637124955654,
      "p95_latency_ms": 3476.5900497324765,
      "input_tokens": 392,
      "output_tokens": 351,
      "reasoning_tokens": 261,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a2ecc5e6-c066-4fde-a521-bf6b199f9ad0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 6.722e-05,
      "cost_per_success_usd": 2.2406666666666667e-05,
      "octane_score": 82.47545373400773,
      "ci_low": 65.65064865883963,
      "ci_high": 111.67076167076168,
      "ci_metric": "octane_score",
      "median_latency_ms": 4213.715208694339,
      "p95_latency_ms": 5075.718371104448,
      "input_tokens": 352,
      "output_tokens": 32,
      "reasoning_tokens": 185,
      "undefined_reason": null
    },
    {
      "snapshot_id": "24651b10-66c2-4dfc-89cb-9f981977f8b3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 5.014e-05,
      "cost_per_success_usd": 1.671333333333333e-05,
      "octane_score": 71.12086158755486,
      "ci_low": 61.81851445298208,
      "ci_high": 83.59750240153697,
      "ci_metric": "octane_score",
      "median_latency_ms": 1677.5854155421257,
      "p95_latency_ms": 2002.6459160260856,
      "input_tokens": 422,
      "output_tokens": 12,
      "reasoning_tokens": 79,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3f14476e-af7c-433f-a1ac-a57d3468adb6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 9.034e-05,
      "cost_per_success_usd": 3.0113333333333335e-05,
      "octane_score": 83.04184193048484,
      "ci_low": 68.07588075880759,
      "ci_high": 103.67606412382534,
      "ci_metric": "octane_score",
      "median_latency_ms": 1321.8310419470072,
      "p95_latency_ms": 6172.737179417162,
      "input_tokens": 623,
      "output_tokens": 35,
      "reasoning_tokens": 190,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e02dc80b-58af-4625-9a5c-b6326370fa33",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 8.93e-05,
      "cost_per_success_usd": 2.9766666666666667e-05,
      "octane_score": 113.88577827547591,
      "ci_low": 95.4301075268817,
      "ci_high": 130.52009456264773,
      "ci_metric": "octane_score",
      "median_latency_ms": 2239.4621670246124,
      "p95_latency_ms": 2589.9802542291577,
      "input_tokens": 565,
      "output_tokens": 47,
      "reasoning_tokens": 198,
      "undefined_reason": null
    },
    {
      "snapshot_id": "05e11eb8-3b38-4352-b6a5-92a2ecf6724e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00039498000000000007,
      "cost_per_success_usd": 0.00013166000000000002,
      "octane_score": 14.036153729302745,
      "ci_low": 13.198046994202015,
      "ci_high": 14.893452150329326,
      "ci_metric": "octane_score",
      "median_latency_ms": 1965.43029230088,
      "p95_latency_ms": 4657.04011619091,
      "input_tokens": 510,
      "output_tokens": 31,
      "reasoning_tokens": 168,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9a33ea76-c2d8-4ff0-8b49-6cba73a10703",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00098919,
      "cost_per_success_usd": 0.00032973000000000003,
      "octane_score": 45.731355957904945,
      "ci_low": 28.04719012525399,
      "ci_high": 70.86017430845017,
      "ci_metric": "octane_score",
      "median_latency_ms": 5163.5112492367625,
      "p95_latency_ms": 5435.021524690092,
      "input_tokens": 392,
      "output_tokens": 36,
      "reasoning_tokens": 905,
      "undefined_reason": null
    },
    {
      "snapshot_id": "afc28424-7f96-47da-97d0-06dbd9be68f7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00047806499999999996,
      "cost_per_success_usd": 0.000159355,
      "octane_score": 21.27325782058925,
      "ci_low": 17.48297825609489,
      "ci_high": 25.746199480904703,
      "ci_metric": "octane_score",
      "median_latency_ms": 1714.9527072906494,
      "p95_latency_ms": 2571.597833652049,
      "input_tokens": 565,
      "output_tokens": 21,
      "reasoning_tokens": 246,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4517465a-0504-4248-97ca-e386c851007b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000552015,
      "cost_per_success_usd": 0.000184005,
      "octane_score": 13.590210410948979,
      "ci_low": 12.371400506643232,
      "ci_high": 14.863356703391043,
      "ci_metric": "octane_score",
      "median_latency_ms": 2395.2027494087815,
      "p95_latency_ms": 5622.560037858783,
      "input_tokens": 623,
      "output_tokens": 6,
      "reasoning_tokens": 317,
      "undefined_reason": null
    },
    {
      "snapshot_id": "11ce7868-51b0-412e-8c5d-fffed155c07f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000203203,
      "cost_per_success_usd": 6.773433333333333e-05,
      "octane_score": 17.548953509544642,
      "ci_low": 10.801934939865973,
      "ci_high": 28.176310277258494,
      "ci_metric": "octane_score",
      "median_latency_ms": 3851.9775830209255,
      "p95_latency_ms": 4411.050833016634,
      "input_tokens": 501,
      "output_tokens": 15,
      "reasoning_tokens": 95,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4a7e45cc-2e36-4281-9fd8-0492b174d24e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00010829999999999999,
      "cost_per_success_usd": 3.6099999999999997e-05,
      "octane_score": 32.92705447830102,
      "ci_low": 30.415512465373965,
      "ci_high": 35.59556786703602,
      "ci_metric": "octane_score",
      "median_latency_ms": 592.5275841727853,
      "p95_latency_ms": 627.2310202941298,
      "input_tokens": 261,
      "output_tokens": 12,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1f80ee8e-03a9-43f3-8167-78c3b3081d80",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.000231,
      "cost_per_success_usd": 7.7e-05,
      "octane_score": 23.999999999999996,
      "ci_low": 22.467532467532468,
      "ci_high": 25.35064935064935,
      "ci_metric": "octane_score",
      "median_latency_ms": 675.503209233284,
      "p95_latency_ms": 700.4167081788182,
      "input_tokens": 270,
      "output_tokens": 60,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "602e7e4c-698f-43bc-9d6d-054aaeca8629",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00016739999999999998,
      "cost_per_success_usd": 5.5799999999999994e-05,
      "octane_score": 60.752688172043015,
      "ci_low": 49.9283154121864,
      "ci_high": 66.77419354838709,
      "ci_metric": "octane_score",
      "median_latency_ms": 583.4970837458968,
      "p95_latency_ms": 638.8976700603961,
      "input_tokens": 483,
      "output_tokens": 9,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9e2de42c-617b-4e38-82f2-d5f17b3c4987",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00016169999999999997,
      "cost_per_success_usd": 5.389999999999999e-05,
      "octane_score": 46.39455782312926,
      "ci_low": 44.397031539888694,
      "ci_high": 49.758812615955485,
      "ci_metric": "octane_score",
      "median_latency_ms": 581.179041415453,
      "p95_latency_ms": 715.2518540620803,
      "input_tokens": 489,
      "output_tokens": 6,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "67de4340-c72d-49a3-b50e-0bff51ac6acb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0002361,
      "cost_per_success_usd": 7.87e-05,
      "octane_score": 191.6010165184244,
      "ci_low": 117.50952986022871,
      "ci_high": 297.01397712833545,
      "ci_metric": "octane_score",
      "median_latency_ms": 754.093499854207,
      "p95_latency_ms": 965.5544245615602,
      "input_tokens": 237,
      "output_tokens": 66,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "abc5dc43-51e8-43a8-b945-b917a399dfad",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.010359,
      "cost_per_success_usd": 0.003453,
      "octane_score": 0.7242011777198571,
      "ci_low": 0.6708321762473388,
      "ci_high": 0.7857791225416035,
      "ci_metric": "octane_score",
      "median_latency_ms": 2683.442666195333,
      "p95_latency_ms": 4166.606929153204,
      "input_tokens": 491,
      "output_tokens": 6,
      "reasoning_tokens": 1277,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4c09eda2-82ff-4c6d-b434-89f9b093b6f4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0111435,
      "cost_per_success_usd": 0.0037145000000000004,
      "octane_score": 0.9126396554044958,
      "ci_low": 0.7610868407667881,
      "ci_high": 1.083796702760706,
      "ci_metric": "octane_score",
      "median_latency_ms": 2984.6404157578945,
      "p95_latency_ms": 3214.9117913097143,
      "input_tokens": 484,
      "output_tokens": 9,
      "reasoning_tokens": 1380,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3802a17f-0ac1-4154-b7bf-d1ea28a9f77a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.010017,
      "cost_per_success_usd": 0.003339,
      "octane_score": 4.51602276130578,
      "ci_low": 2.827269948028126,
      "ci_high": 6.860874669797477,
      "ci_metric": "octane_score",
      "median_latency_ms": 2594.638915732503,
      "p95_latency_ms": 3544.54744188115,
      "input_tokens": 238,
      "output_tokens": 41,
      "reasoning_tokens": 1247,
      "undefined_reason": null
    },
    {
      "snapshot_id": "24bf70bb-c34b-42b0-8ca7-b4a8b487cb7d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0073425,
      "cost_per_success_usd": 0.0024475,
      "octane_score": 0.7550561797752808,
      "ci_low": 0.6929687499999999,
      "ci_high": 0.8268456375838925,
      "ci_metric": "octane_score",
      "median_latency_ms": 2493.345083668828,
      "p95_latency_ms": 2895.7169716246426,
      "input_tokens": 270,
      "output_tokens": 81,
      "reasoning_tokens": 844,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8fac468e-5169-44bf-99c8-2d254273a45d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007363499999999999,
      "cost_per_success_usd": 0.0024544999999999997,
      "octane_score": 0.4842805730970327,
      "ci_low": 0.43792120269569734,
      "ci_high": 0.5353016688061617,
      "ci_metric": "octane_score",
      "median_latency_ms": 2198.682291433215,
      "p95_latency_ms": 2388.201853726059,
      "input_tokens": 264,
      "output_tokens": 12,
      "reasoning_tokens": 917,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8baf3f07-7889-434f-ab52-d4d2242139cd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001242375,
      "cost_per_success_usd": 0.000414125,
      "octane_score": 8.18593419861153,
      "ci_low": 6.8313442501619015,
      "ci_high": 9.41126641253706,
      "ci_metric": "octane_score",
      "median_latency_ms": 2060.1188326254487,
      "p95_latency_ms": 2465.9627714194357,
      "input_tokens": 483,
      "output_tokens": 9,
      "reasoning_tokens": 557,
      "undefined_reason": null
    },
    {
      "snapshot_id": "dd2e2cdf-53ee-47de-b7fb-a0280a850ce0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0013916249999999998,
      "cost_per_success_usd": 0.00046387499999999993,
      "octane_score": 2.5624719302973147,
      "ci_low": 2.1865243063981237,
      "ci_high": 3.119175627240143,
      "ci_metric": "octane_score",
      "median_latency_ms": 2083.554374985397,
      "p95_latency_ms": 2547.922387812286,
      "input_tokens": 261,
      "output_tokens": 12,
      "reasoning_tokens": 678,
      "undefined_reason": null
    },
    {
      "snapshot_id": "36759a14-0bd5-4a86-9add-19ceaef718be",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0013012499999999999,
      "cost_per_success_usd": 0.00043374999999999995,
      "octane_score": 4.260518731988473,
      "ci_low": 3.986516853932584,
      "ci_high": 4.5461426491994175,
      "ci_metric": "octane_score",
      "median_latency_ms": 2505.937167443335,
      "p95_latency_ms": 2538.999753817916,
      "input_tokens": 270,
      "output_tokens": 74,
      "reasoning_tokens": 566,
      "undefined_reason": null
    },
    {
      "snapshot_id": "619eddeb-3b2a-45ae-9581-16822523f398",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00162525,
      "cost_per_success_usd": 0.00054175,
      "octane_score": 4.61590524534687,
      "ci_low": 4.333108057370279,
      "ci_high": 4.944918183913344,
      "ci_metric": "octane_score",
      "median_latency_ms": 2290.537125431001,
      "p95_latency_ms": 2583.5955372080202,
      "input_tokens": 489,
      "output_tokens": 6,
      "reasoning_tokens": 763,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8ec06016-6931-432d-8b81-b0973ec32d11",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0019920000000000003,
      "cost_per_success_usd": 0.0006640000000000001,
      "octane_score": 22.709337349397586,
      "ci_low": 13.810425116418807,
      "ci_high": 35.9210347174949,
      "ci_metric": "octane_score",
      "median_latency_ms": 2771.310333162546,
      "p95_latency_ms": 3303.6310829222202,
      "input_tokens": 237,
      "output_tokens": 52,
      "reasoning_tokens": 963,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f8b02f0c-b42e-4cc0-bb3c-69e65f4800f6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00079706,
      "cost_per_success_usd": 0.0002656866666666667,
      "octane_score": 56.754823978119575,
      "ci_low": 35.232265765880165,
      "ci_high": 87.80332056194126,
      "ci_metric": "octane_score",
      "median_latency_ms": 7254.062082618475,
      "p95_latency_ms": 14742.698533367364,
      "input_tokens": 231,
      "output_tokens": 705,
      "reasoning_tokens": 1269,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7911ca0f-bf49-48d2-b7ce-57ed1b5f3e4f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0006868,
      "cost_per_success_usd": 0.00022893333333333333,
      "octane_score": 14.807804309842748,
      "ci_low": 11.330213903743314,
      "ci_high": 19.115622849277354,
      "ci_metric": "octane_score",
      "median_latency_ms": 7832.034250721335,
      "p95_latency_ms": 37510.82766270265,
      "input_tokens": 480,
      "output_tokens": 280,
      "reasoning_tokens": 1397,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b41ab254-46cf-4b35-bf4c-79dab0f99bc7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00057532,
      "cost_per_success_usd": 0.00019177333333333336,
      "octane_score": 13.039699645414725,
      "ci_low": 12.256168014680894,
      "ci_high": 13.98526037683376,
      "ci_metric": "octane_score",
      "median_latency_ms": 39568.081833422184,
      "p95_latency_ms": 70346.02777026594,
      "input_tokens": 462,
      "output_tokens": 0,
      "reasoning_tokens": 1369,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b24db404-a3d6-487c-a03e-a94c481209d6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00042918000000000003,
      "cost_per_success_usd": 0.00014306,
      "octane_score": 12.917656927163426,
      "ci_low": 10.653753026634382,
      "ci_high": 15.366705471478461,
      "ci_metric": "octane_score",
      "median_latency_ms": 6143.086375668645,
      "p95_latency_ms": 9482.275411859155,
      "input_tokens": 273,
      "output_tokens": 65,
      "reasoning_tokens": 967,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b6e598cf-de04-47d6-83f4-0da98ecf342e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00048872,
      "cost_per_success_usd": 0.00016290666666666666,
      "octane_score": 7.296611556719594,
      "ci_low": 6.678921568627452,
      "ci_high": 7.959365456396335,
      "ci_metric": "octane_score",
      "median_latency_ms": 7173.093416728079,
      "p95_latency_ms": 35700.60855392367,
      "input_tokens": 252,
      "output_tokens": 0,
      "reasoning_tokens": 1184,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c69e3f53-301b-4cf4-af03-ca7d1904dafc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0030954,
      "cost_per_success_usd": 0.0010318,
      "octane_score": 3.285520449699554,
      "ci_low": 2.2548272357723578,
      "ci_high": 7.965118112719588,
      "ci_metric": "octane_score",
      "median_latency_ms": 8403.408999554813,
      "p95_latency_ms": 8508.880637120456,
      "input_tokens": 487,
      "output_tokens": 103,
      "reasoning_tokens": 1357,
      "undefined_reason": null
    },
    {
      "snapshot_id": "69c834ba-5805-406f-a3ac-cd22597a82bb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00238068,
      "cost_per_success_usd": 0.00119034,
      "octane_score": 0.9985942391809622,
      "ci_low": 0.4448965333004542,
      "ci_high": 1.8697008140395408,
      "ci_metric": "octane_score",
      "median_latency_ms": 3057.3835410177708,
      "p95_latency_ms": 5049.740767292677,
      "input_tokens": 259,
      "output_tokens": 4,
      "reasoning_tokens": 1155,
      "undefined_reason": null
    },
    {
      "snapshot_id": "93c5255d-bee6-417a-ae3d-97110888cc05",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0026367599999999997,
      "cost_per_success_usd": 0.0008789199999999999,
      "octane_score": 2.1025804396304557,
      "ci_low": 1.653457325020255,
      "ci_high": 2.570480928689884,
      "ci_metric": "octane_score",
      "median_latency_ms": 10150.42091626674,
      "p95_latency_ms": 10645.737642236054,
      "input_tokens": 273,
      "output_tokens": 59,
      "reasoning_tokens": 1229,
      "undefined_reason": null
    },
    {
      "snapshot_id": "314c6a4f-8d69-4f8b-a3d4-a927f6537f02",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0023773199999999996,
      "cost_per_success_usd": 0.0007924399999999998,
      "octane_score": 19.02856998637121,
      "ci_low": 8.486921792647397,
      "ci_high": 54.512396694214885,
      "ci_metric": "octane_score",
      "median_latency_ms": 5185.490959323943,
      "p95_latency_ms": 13170.6027707085,
      "input_tokens": 231,
      "output_tokens": 40,
      "reasoning_tokens": 1142,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4bdb6798-62c4-42f0-8c67-767eff2af489",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0027486,
      "cost_per_success_usd": 0.0009161999999999999,
      "octane_score": 2.7293895073855783,
      "ci_low": 2.3299609788872524,
      "ci_high": 3.4922665510263062,
      "ci_metric": "octane_score",
      "median_latency_ms": 6714.309125207365,
      "p95_latency_ms": 9828.888125251979,
      "input_tokens": 469,
      "output_tokens": 11,
      "reasoning_tokens": 1274,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9ab5bec5-a53a-4cc0-9044-a360275330c8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0015320759999999998,
      "cost_per_success_usd": 0.000510692,
      "octane_score": 3.6186194418553645,
      "ci_low": 3.4111576798764567,
      "ci_high": 3.8183255887261893,
      "ci_metric": "octane_score",
      "median_latency_ms": 2322.252250276506,
      "p95_latency_ms": 2339.931249804795,
      "input_tokens": 294,
      "output_tokens": 56,
      "reasoning_tokens": 420,
      "undefined_reason": null
    },
    {
      "snapshot_id": "22967d88-f11d-472a-9ebf-c09951615a7a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0022238976,
      "cost_per_success_usd": 0.0007412992,
      "octane_score": 20.34131427634078,
      "ci_low": 12.860033663029293,
      "ci_high": 30.793368919709337,
      "ci_metric": "octane_score",
      "median_latency_ms": 3309.000415727496,
      "p95_latency_ms": 3374.5284038595855,
      "input_tokens": 252,
      "output_tokens": 118,
      "reasoning_tokens": 572,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1a07d3fd-b2df-4c2c-b9a7-c0704a2cbbe6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0017553876,
      "cost_per_success_usd": 0.0005851292,
      "octane_score": 5.793592252787931,
      "ci_low": 4.853532256464424,
      "ci_high": 6.522079617039452,
      "ci_metric": "octane_score",
      "median_latency_ms": 1692.3546250909567,
      "p95_latency_ms": 2930.085512157529,
      "input_tokens": 501,
      "output_tokens": 34,
      "reasoning_tokens": 453,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f312083d-d23e-4f8d-85b7-e58ad3fdd0ff",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016820268000000001,
      "cost_per_success_usd": 0.0005606756,
      "octane_score": 4.460095403949568,
      "ci_low": 2.514189991794531,
      "ci_high": 7.564277341883492,
      "ci_metric": "octane_score",
      "median_latency_ms": 3140.707666054368,
      "p95_latency_ms": 3494.9136540293694,
      "input_tokens": 483,
      "output_tokens": 12,
      "reasoning_tokens": 495,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e4b03336-3081-4e80-a1b1-4ea90f58e278",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0024387084,
      "cost_per_success_usd": 0.0008129028000000001,
      "octane_score": 1.4622494431888617,
      "ci_low": 1.239178042015294,
      "ci_high": 1.7248900951323078,
      "ci_metric": "octane_score",
      "median_latency_ms": 3037.91725076735,
      "p95_latency_ms": 3278.5090373829007,
      "input_tokens": 273,
      "output_tokens": 10,
      "reasoning_tokens": 726,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bedf58dd-a62d-4a20-9b86-0c31566ca8ba",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": 3,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.0036843600000000002,
      "cost_per_success_usd": 0.0036843600000000002,
      "octane_score": 0.6787248441158482,
      "ci_low": 0.6344395535898755,
      "ci_high": 2.3957335377147597,
      "ci_metric": "octane_score",
      "median_latency_ms": 6204.442041926086,
      "p95_latency_ms": 6864.966404624283,
      "input_tokens": 483,
      "output_tokens": 128,
      "reasoning_tokens": 622,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6f3faa1e-0b5a-4605-bc00-b2cf052e3c7f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00266948,
      "cost_per_success_usd": 0.0008898266666666667,
      "octane_score": 1.3358406880740819,
      "ci_low": 1.159845125149314,
      "ci_high": 1.5795444232688995,
      "ci_metric": "octane_score",
      "median_latency_ms": 4256.065499968827,
      "p95_latency_ms": 4982.910562586039,
      "input_tokens": 273,
      "output_tokens": 15,
      "reasoning_tokens": 538,
      "undefined_reason": null
    },
    {
      "snapshot_id": "58bfcc5b-a7bc-4fcf-a8a0-51cce292d6e1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00249928,
      "cost_per_success_usd": 0.0008330933333333334,
      "octane_score": 18.10001280368746,
      "ci_low": 11.434600547330938,
      "ci_high": 27.716514892809716,
      "ci_metric": "octane_score",
      "median_latency_ms": 3531.354082748294,
      "p95_latency_ms": 3853.5956712439656,
      "input_tokens": 252,
      "output_tokens": 36,
      "reasoning_tokens": 485,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8fb97f42-5dcd-42c8-a247-7ee80d610af6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0035128800000000003,
      "cost_per_success_usd": 0.0011709600000000002,
      "octane_score": 1.57819225251076,
      "ci_low": 1.4325581395348834,
      "ci_high": 1.7292881307133086,
      "ci_metric": "octane_score",
      "median_latency_ms": 4700.40912553668,
      "p95_latency_ms": 5223.060049954801,
      "input_tokens": 294,
      "output_tokens": 206,
      "reasoning_tokens": 532,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1dc5a163-4618-47c4-a7ff-74d8103d537d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0017911600000000002,
      "cost_per_success_usd": 0.0005970533333333334,
      "octane_score": 5.677884722749502,
      "ci_low": 4.666249804595905,
      "ci_high": 6.734802744841296,
      "ci_metric": "octane_score",
      "median_latency_ms": 3216.7828334495425,
      "p95_latency_ms": 3254.262807685882,
      "input_tokens": 501,
      "output_tokens": 15,
      "reasoning_tokens": 299,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2df2813f-6275-42c4-ae65-e8c760ca0f7f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00042975000000000006,
      "cost_per_success_usd": 0.00014325,
      "octane_score": 12.900523560209422,
      "ci_low": 12.07678883071553,
      "ci_high": 13.704099821746878,
      "ci_metric": "octane_score",
      "median_latency_ms": 770.1250845566392,
      "p95_latency_ms": 995.0646334327757,
      "input_tokens": 273,
      "output_tokens": 50,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e52c5bc0-cd99-443c-ab89-7e38c9f726b3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0002835,
      "cost_per_success_usd": 9.45e-05,
      "octane_score": 12.578483245149911,
      "ci_low": 11.619047619047619,
      "ci_high": 13.597883597883598,
      "ci_metric": "octane_score",
      "median_latency_ms": 589.1492506489158,
      "p95_latency_ms": 592.8679373115301,
      "input_tokens": 252,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "76cc794f-8dad-489d-8968-fee5b6304c3c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00043875,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 630.9425411745906,
      "p95_latency_ms": 680.5223793722689,
      "input_tokens": 225,
      "output_tokens": 60,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "c7a0a775-bc3f-4d1b-ac1d-770d1344cb9e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00045674999999999996,
      "cost_per_success_usd": 0.00015225,
      "octane_score": 22.26600985221675,
      "ci_low": 18.29885057471265,
      "ci_high": 24.472906403940883,
      "ci_metric": "octane_score",
      "median_latency_ms": 653.1942496076226,
      "p95_latency_ms": 661.8714874610305,
      "input_tokens": 483,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a86f760c-c7d4-4518-921f-b63621b0918a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00042525,
      "cost_per_success_usd": 0.00014175,
      "octane_score": 17.641387419165195,
      "ci_low": 16.88183421516755,
      "ci_high": 18.92063492063492,
      "ci_metric": "octane_score",
      "median_latency_ms": 733.6895000189543,
      "p95_latency_ms": 907.8120132908225,
      "input_tokens": 459,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c5f91341-9f60-4740-96ac-a732a4955503",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 7.560000000000001e-05,
      "cost_per_success_usd": 2.5200000000000003e-05,
      "octane_score": 47.169312169312164,
      "ci_low": 43.57142857142857,
      "ci_high": 50.992063492063494,
      "ci_metric": "octane_score",
      "median_latency_ms": 678.3937904983759,
      "p95_latency_ms": 822.2873287275434,
      "input_tokens": 252,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "79ded3e9-4318-4526-8c3f-d485c630c738",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00018539999999999995,
      "cost_per_success_usd": 6.179999999999998e-05,
      "octane_score": 29.902912621359228,
      "ci_low": 27.418397626112764,
      "ci_high": 32.57627118644068,
      "ci_metric": "octane_score",
      "median_latency_ms": 1004.5221671462059,
      "p95_latency_ms": 1010.7934786006808,
      "input_tokens": 273,
      "output_tokens": 54,
      "reasoning_tokens": 55,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2efb8d94-881c-4297-a52e-39b5309e28ff",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00044459999999999996,
      "cost_per_success_usd": 0.0001482,
      "octane_score": 101.74763832658569,
      "ci_low": 61.21605667060213,
      "ci_high": 157.72604588394063,
      "ci_metric": "octane_score",
      "median_latency_ms": 2591.2002502009273,
      "p95_latency_ms": 2815.7294370234013,
      "input_tokens": 225,
      "output_tokens": 45,
      "reasoning_tokens": 288,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cf04c59e-f6a8-4830-a266-70670fb90d50",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00011339999999999999,
      "cost_per_success_usd": 3.78e-05,
      "octane_score": 66.1552028218695,
      "ci_low": 63.30687830687832,
      "ci_high": 70.95238095238095,
      "ci_metric": "octane_score",
      "median_latency_ms": 699.047208763659,
      "p95_latency_ms": 840.4258457943797,
      "input_tokens": 459,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fbe95b16-4239-46bb-8ac7-9cc402af2bcf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001218,
      "cost_per_success_usd": 4.06e-05,
      "octane_score": 83.49753694581281,
      "ci_low": 68.62068965517244,
      "ci_high": 91.7733990147783,
      "ci_metric": "octane_score",
      "median_latency_ms": 888.690416701138,
      "p95_latency_ms": 1211.7894051596522,
      "input_tokens": 483,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3a75fa4a-6d5f-4c35-9252-b5b7cedcb47b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00083016,
      "cost_per_success_usd": 0.00027672,
      "octane_score": 4.295557482894863,
      "ci_low": 2.7299237326155223,
      "ci_high": 6.605766184391172,
      "ci_metric": "octane_score",
      "median_latency_ms": 1954.785542562604,
      "p95_latency_ms": 2196.6351544484496,
      "input_tokens": 5481,
      "output_tokens": 114,
      "reasoning_tokens": 75,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fd4e7cf5-a734-41ad-a43a-8652baa3021e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00126626,
      "cost_per_success_usd": 0.00042208666666666665,
      "octane_score": 5.924533666071739,
      "ci_low": 4.643960855009014,
      "ci_high": 10.33518779927482,
      "ci_metric": "octane_score",
      "median_latency_ms": 1965.8115422353148,
      "p95_latency_ms": 2250.783341936767,
      "input_tokens": 6313,
      "output_tokens": 107,
      "reasoning_tokens": 111,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ee008f8e-e1f2-477d-ad03-81d8c452a4a4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0027814,
      "cost_per_success_usd": 0.0009271333333333333,
      "octane_score": 16.264111598475587,
      "ci_low": 10.048533140166606,
      "ci_high": 24.95387885671458,
      "ci_metric": "octane_score",
      "median_latency_ms": 4537.797832861543,
      "p95_latency_ms": 5769.679507706313,
      "input_tokens": 6251,
      "output_tokens": 186,
      "reasoning_tokens": 1090,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c84fb279-95ca-4f1d-acda-3ec21ce34dec",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00138092,
      "cost_per_success_usd": 0.0004603066666666667,
      "octane_score": 7.3646554470932415,
      "ci_low": 5.460246095598674,
      "ci_high": 13.022433927473877,
      "ci_metric": "octane_score",
      "median_latency_ms": 2264.756958000362,
      "p95_latency_ms": 2554.186121374368,
      "input_tokens": 6430,
      "output_tokens": 116,
      "reasoning_tokens": 180,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d71d624a-4dea-46eb-9501-b8706b28a6fc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016676,
      "cost_per_success_usd": 0.0005558666666666667,
      "octane_score": 3.3245382585751977,
      "ci_low": 3.1309565831274266,
      "ci_high": 3.511633485248261,
      "ci_metric": "octane_score",
      "median_latency_ms": 2417.540915310383,
      "p95_latency_ms": 3038.1698538549244,
      "input_tokens": 5746,
      "output_tokens": 213,
      "reasoning_tokens": 219,
      "undefined_reason": null
    },
    {
      "snapshot_id": "745d92b4-9377-41af-83bc-5738c4350e55",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00312,
      "cost_per_success_usd": 0.0010400000000000001,
      "octane_score": 14.499038461538458,
      "ci_low": 8.892307692307691,
      "ci_high": 21.98666666666667,
      "ci_metric": "octane_score",
      "median_latency_ms": 3458.0526668578386,
      "p95_latency_ms": 3585.5375550687313,
      "input_tokens": 225,
      "output_tokens": 45,
      "reasoning_tokens": 222,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ff1c1b08-d1ec-4b70-b1fa-d7699f5216f4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001176,
      "cost_per_success_usd": 0.000392,
      "octane_score": 8.64795918367347,
      "ci_low": 7.107142857142859,
      "ci_high": 9.505102040816325,
      "ci_metric": "octane_score",
      "median_latency_ms": 874.6698750182986,
      "p95_latency_ms": 902.2923757322133,
      "input_tokens": 483,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "247603bc-1baa-4ea7-be63-15b655212cda",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001026,
      "cost_per_success_usd": 0.000342,
      "octane_score": 5.403508771929824,
      "ci_low": 5.058479532163743,
      "ci_high": 5.707602339181286,
      "ci_metric": "octane_score",
      "median_latency_ms": 1842.863041907549,
      "p95_latency_ms": 2403.4391780383885,
      "input_tokens": 273,
      "output_tokens": 48,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "58b2d4ac-5abe-4217-b051-8c7900f7b945",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000714,
      "cost_per_success_usd": 0.000238,
      "octane_score": 4.9943977591036415,
      "ci_low": 4.61344537815126,
      "ci_high": 5.399159663865547,
      "ci_metric": "octane_score",
      "median_latency_ms": 1671.3009579107165,
      "p95_latency_ms": 1682.5852719135582,
      "input_tokens": 252,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "47b8c252-166c-45b1-9969-7a0cf8952c65",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001098,
      "cost_per_success_usd": 0.000366,
      "octane_score": 6.832422586520947,
      "ci_low": 6.538251366120219,
      "ci_high": 7.327868852459016,
      "ci_metric": "octane_score",
      "median_latency_ms": 1067.6153330132365,
      "p95_latency_ms": 1088.4877203032374,
      "input_tokens": 459,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "65e763be-4b5d-4959-831e-6c311b5d6fcf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.021543,
      "cost_per_success_usd": 0.007181,
      "octane_score": 0.3482337650280834,
      "ci_low": 0.21881533101045297,
      "ci_high": 0.5161482659246746,
      "ci_metric": "octane_score",
      "median_latency_ms": 2851.9444577395916,
      "p95_latency_ms": 3163.0047458224,
      "input_tokens": 6294,
      "output_tokens": 99,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ca6b1217-7b26-4815-92d4-566bf46bf480",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0566,
      "cost_per_success_usd": 0.018866666666666667,
      "octane_score": 0.7992402826855123,
      "ci_low": 0.504991477964451,
      "ci_high": 1.236663433559651,
      "ci_metric": "octane_score",
      "median_latency_ms": 5990.899875760078,
      "p95_latency_ms": 6272.759624850005,
      "input_tokens": 6076,
      "output_tokens": 180,
      "reasoning_tokens": 694,
      "undefined_reason": null
    },
    {
      "snapshot_id": "57ad71bc-639b-4258-bde9-3f62ef99b8cb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.021237,
      "cost_per_success_usd": 0.007078999999999999,
      "octane_score": 0.261053821161181,
      "ci_low": 0.1666370896184561,
      "ci_high": 0.3856727199759205,
      "ci_metric": "octane_score",
      "median_latency_ms": 3013.7565825134516,
      "p95_latency_ms": 3298.765771370381,
      "input_tokens": 5628,
      "output_tokens": 189,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8003f17e-3f52-48e4-b262-c22db8d181a7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0244845,
      "cost_per_success_usd": 0.0081615,
      "octane_score": 0.14564316200044924,
      "ci_low": 0.11026268559308207,
      "ci_high": 0.2931721955029391,
      "ci_metric": "octane_score",
      "median_latency_ms": 2940.087500028312,
      "p95_latency_ms": 3404.4661629013717,
      "input_tokens": 5481,
      "output_tokens": 108,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d2b1b597-3541-45ca-ab25-32a6d4de004f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.022251,
      "cost_per_success_usd": 0.007417000000000001,
      "octane_score": 0.4570581097478764,
      "ci_low": 0.2906321719173796,
      "ci_high": 0.705665650406504,
      "ci_metric": "octane_score",
      "median_latency_ms": 2273.716832511127,
      "p95_latency_ms": 2280.5422078818083,
      "input_tokens": 6405,
      "output_tokens": 108,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6c603b96-071b-4c50-9741-d4364d88a9a1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001218,
      "cost_per_success_usd": 0.00040599999999999995,
      "octane_score": 8.349753694581281,
      "ci_low": 6.862068965517244,
      "ci_high": 9.177339901477833,
      "ci_metric": "octane_score",
      "median_latency_ms": 680.7530000805855,
      "p95_latency_ms": 700.0725878402591,
      "input_tokens": 483,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "74a5d0a0-f45a-4f4d-b515-ff6625b4f04b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001134,
      "cost_per_success_usd": 0.000378,
      "octane_score": 6.6155202821869485,
      "ci_low": 6.330687830687831,
      "ci_high": 7.095238095238095,
      "ci_metric": "octane_score",
      "median_latency_ms": 875.0442089512944,
      "p95_latency_ms": 1710.241770930588,
      "input_tokens": 459,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "31175441-4c45-4702-b7ad-a6996524e09d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0007559999999999999,
      "cost_per_success_usd": 0.000252,
      "octane_score": 4.716931216931217,
      "ci_low": 4.357142857142858,
      "ci_high": 5.09920634920635,
      "ci_metric": "octane_score",
      "median_latency_ms": 808.6224999278784,
      "p95_latency_ms": 1140.5078502371907,
      "input_tokens": 252,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a19b0f91-ecd2-4dee-b066-13e9e39872ae",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0038459999999999996,
      "cost_per_success_usd": 0.001282,
      "octane_score": 11.762090483619344,
      "ci_low": 7.3984,
      "ci_high": 17.789193302891935,
      "ci_metric": "octane_score",
      "median_latency_ms": 2625.1312075182796,
      "p95_latency_ms": 2665.6458330340683,
      "input_tokens": 225,
      "output_tokens": 46,
      "reasoning_tokens": 237,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d8102e5d-16c8-4210-b59d-05e2addba236",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001122,
      "cost_per_success_usd": 0.000374,
      "octane_score": 4.941176470588235,
      "ci_low": 4.625668449197861,
      "ci_high": 5.219251336898396,
      "ci_metric": "octane_score",
      "median_latency_ms": 951.0467909276485,
      "p95_latency_ms": 1007.7440165914595,
      "input_tokens": 273,
      "output_tokens": 48,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0e04e617-4cae-4c14-96a7-cae417e0ca2e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0040128,
      "cost_per_success_usd": 0.0013376000000000002,
      "octane_score": 0.8886562998405102,
      "ci_low": 0.5705578996718237,
      "ci_high": 1.358453686610924,
      "ci_metric": "octane_score",
      "median_latency_ms": 2113.235999830067,
      "p95_latency_ms": 2441.919411998242,
      "input_tokens": 5481,
      "output_tokens": 114,
      "reasoning_tokens": 52,
      "undefined_reason": null
    },
    {
      "snapshot_id": "90022cf3-08e6-4321-b6e0-64a5ebf9508d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0047094,
      "cost_per_success_usd": 0.0015697999999999999,
      "octane_score": 1.1772200280290483,
      "ci_low": 0.7811980033277869,
      "ci_high": 1.6731957865413076,
      "ci_metric": "octane_score",
      "median_latency_ms": 1884.2269582673907,
      "p95_latency_ms": 2050.2491324208677,
      "input_tokens": 5628,
      "output_tokens": 195,
      "reasoning_tokens": 71,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6d6a1294-b20f-47ce-a71d-a8bf2c3acd69",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004746600000000001,
      "cost_per_success_usd": 0.0015822000000000004,
      "octane_score": 1.5804997261197484,
      "ci_low": 1.044925124792013,
      "ci_high": 2.249848392965433,
      "ci_metric": "octane_score",
      "median_latency_ms": 2041.3060411810875,
      "p95_latency_ms": 2632.0956282317634,
      "input_tokens": 6294,
      "output_tokens": 105,
      "reasoning_tokens": 67,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c64ac648-92f8-4a69-954f-81203a1286f2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.012191,
      "cost_per_success_usd": 0.004063666666666667,
      "octane_score": 3.710688212615864,
      "ci_low": 2.305084745762712,
      "ci_high": 5.711435087147743,
      "ci_metric": "octane_score",
      "median_latency_ms": 4179.084708914161,
      "p95_latency_ms": 4508.571334555745,
      "input_tokens": 6107,
      "output_tokens": 187,
      "reasoning_tokens": 827,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2fcce9c1-1779-4dd8-9d1b-843af15822a2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0051282,
      "cost_per_success_usd": 0.0017094000000000002,
      "octane_score": 1.9831519831519828,
      "ci_low": 1.228828510938603,
      "ci_high": 3.215815676739608,
      "ci_metric": "octane_score",
      "median_latency_ms": 1979.5916248112917,
      "p95_latency_ms": 2698.8553507253523,
      "input_tokens": 6405,
      "output_tokens": 114,
      "reasoning_tokens": 107,
      "undefined_reason": null
    },
    {
      "snapshot_id": "dfbabf48-523a-4dc4-9bdd-a42717f35344",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 3.566e-05,
      "cost_per_success_usd": 1.1886666666666667e-05,
      "octane_score": 99.99999999999999,
      "ci_low": 1.098e-05,
      "ci_high": 1.285e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 3548.7728333100677,
      "p95_latency_ms": 6166.058258991688,
      "input_tokens": 435,
      "output_tokens": 35,
      "reasoning_tokens": 98,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5dc7afef-875b-4ae6-bc5a-248b2446ac27",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 5.544e-05,
      "cost_per_success_usd": 1.848e-05,
      "octane_score": 100.0,
      "ci_low": 1.73e-05,
      "ci_high": 1.952e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 1784.589291550219,
      "p95_latency_ms": 5819.8411049321285,
      "input_tokens": 454,
      "output_tokens": 98,
      "reasoning_tokens": 148,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e9bbaa64-c1e2-4d69-9963-de8224013440",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 7.502e-05,
      "cost_per_success_usd": 2.5006666666666666e-05,
      "octane_score": 100.0,
      "ci_low": 2.393e-05,
      "ci_high": 2.682e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 4496.205290779471,
      "p95_latency_ms": 5140.495191980152,
      "input_tokens": 642,
      "output_tokens": 33,
      "reasoning_tokens": 295,
      "undefined_reason": null
    },
    {
      "snapshot_id": "33ea70c8-c86f-4918-b400-8a0e48566dac",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001017,
      "cost_per_success_usd": 3.39e-05,
      "octane_score": 100.0,
      "ci_low": 2.7860000000000008e-05,
      "ci_high": 3.725999999999999e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 3007.6647493988276,
      "p95_latency_ms": 3768.754999432713,
      "input_tokens": 653,
      "output_tokens": 65,
      "reasoning_tokens": 418,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0e918d5a-b306-4ec8-bc7d-932d6f816a35",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00045237,
      "cost_per_success_usd": 0.00015078999999999998,
      "octane_score": 100.0,
      "ci_low": 9.248e-05,
      "ci_high": 0.00023375000000000002,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 30283.295709639788,
      "p95_latency_ms": 50940.6286583282,
      "input_tokens": 408,
      "output_tokens": 37,
      "reasoning_tokens": 2552,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ce7fe2be-76e7-429b-98d0-db7dee5bc05c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003568,
      "cost_per_success_usd": 0.0011893333333333333,
      "octane_score": 2.102578475336323,
      "ci_low": 1.882494004796163,
      "ci_high": 2.3253939828080226,
      "ci_metric": "octane_score",
      "median_latency_ms": 2651.9418340176344,
      "p95_latency_ms": 3005.0125082954764,
      "input_tokens": 1059,
      "output_tokens": 6,
      "reasoning_tokens": 417,
      "undefined_reason": null
    },
    {
      "snapshot_id": "41a8aec1-9b53-471f-bc59-11b4b0698444",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.006263599999999999,
      "cost_per_success_usd": 0.002087866666666666,
      "octane_score": 1.6236669008238076,
      "ci_low": 1.293586724969167,
      "ci_high": 2.0604599406528186,
      "ci_metric": "octane_score",
      "median_latency_ms": 4074.253249913454,
      "p95_latency_ms": 5077.695837430655,
      "input_tokens": 1080,
      "output_tokens": 9,
      "reasoning_tokens": 820,
      "undefined_reason": null
    },
    {
      "snapshot_id": "860c5cec-a435-4f06-869c-21f89712200e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0081492,
      "cost_per_success_usd": 0.0027164,
      "octane_score": 5.551097040200265,
      "ci_low": 3.1510817599870347,
      "ci_high": 9.2386703187251,
      "ci_metric": "octane_score",
      "median_latency_ms": 5091.661707498133,
      "p95_latency_ms": 8281.056509073824,
      "input_tokens": 834,
      "output_tokens": 30,
      "reasoning_tokens": 1159,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d5ec1545-6bb4-4c00-807a-7c162d427969",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0029036,
      "cost_per_success_usd": 0.0009678666666666667,
      "octane_score": 1.2281305965008953,
      "ci_low": 1.0825964372677177,
      "ci_high": 1.444660508831632,
      "ci_metric": "octane_score",
      "median_latency_ms": 1929.6990418806672,
      "p95_latency_ms": 2020.4269545152783,
      "input_tokens": 861,
      "output_tokens": 9,
      "reasoning_tokens": 333,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f234b6eb-efc8-4803-b280-d92f03618b0b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0034136,
      "cost_per_success_usd": 0.0011378666666666667,
      "octane_score": 1.6240918678228262,
      "ci_low": 1.4554715798591773,
      "ci_high": 1.8918918918918919,
      "ci_metric": "octane_score",
      "median_latency_ms": 2547.738124616444,
      "p95_latency_ms": 2658.615388162434,
      "input_tokens": 885,
      "output_tokens": 50,
      "reasoning_tokens": 369,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4c64cedd-3acd-4844-86c0-dad55513ea82",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00606,
      "cost_per_success_usd": 0.00202,
      "octane_score": 1.2379537953795379,
      "ci_low": 1.142743979721166,
      "ci_high": 1.3414256198347105,
      "ci_metric": "octane_score",
      "median_latency_ms": 3008.773500099778,
      "p95_latency_ms": 3156.0140999034047,
      "input_tokens": 1059,
      "output_tokens": 6,
      "reasoning_tokens": 811,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d3a05215-9a9b-47ac-b60d-23244cb68c60",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009108,
      "cost_per_success_usd": 0.003036,
      "octane_score": 1.116600790513834,
      "ci_low": 0.9292861907938629,
      "ci_high": 1.24356391314081,
      "ci_metric": "octane_score",
      "median_latency_ms": 4167.800749652088,
      "p95_latency_ms": 4747.591662034392,
      "input_tokens": 1080,
      "output_tokens": 9,
      "reasoning_tokens": 1309,
      "undefined_reason": null
    },
    {
      "snapshot_id": "81435cda-ba4e-48a2-a63d-b3d931c8e2e3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.012468,
      "cost_per_success_usd": 0.004156,
      "octane_score": 3.6282483156881615,
      "ci_low": 2.246841593780369,
      "ci_high": 5.610897743638983,
      "ci_metric": "octane_score",
      "median_latency_ms": 8986.4480830729,
      "p95_latency_ms": 9126.00050875917,
      "input_tokens": 834,
      "output_tokens": 30,
      "reasoning_tokens": 1930,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c7054b4f-a5b9-4a8a-b00f-e6d67721464b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.005202,
      "cost_per_success_usd": 0.001734,
      "octane_score": 1.065743944636678,
      "ci_low": 0.9866518353726363,
      "ci_high": 1.1426872770511294,
      "ci_metric": "octane_score",
      "median_latency_ms": 3928.6232497543097,
      "p95_latency_ms": 4114.208200015128,
      "input_tokens": 885,
      "output_tokens": 45,
      "reasoning_tokens": 687,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d699a1fc-c92d-4dae-bec8-a2252d17375f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0040680000000000004,
      "cost_per_success_usd": 0.0013560000000000002,
      "octane_score": 0.8765978367748277,
      "ci_low": 0.7931924882629108,
      "ci_high": 0.9688356561551386,
      "ci_metric": "octane_score",
      "median_latency_ms": 3493.6769995838404,
      "p95_latency_ms": 3598.4895748086274,
      "input_tokens": 861,
      "output_tokens": 9,
      "reasoning_tokens": 510,
      "undefined_reason": null
    },
    {
      "snapshot_id": "78edecb4-ad4c-4335-9534-bf9b425d66b3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00705477,
      "cost_per_success_usd": 0.0023515899999999998,
      "octane_score": 6.412257238719334,
      "ci_low": 3.9326583290454544,
      "ci_high": 9.632343245203403,
      "ci_metric": "octane_score",
      "median_latency_ms": 14866.655374877155,
      "p95_latency_ms": 17139.2898125574,
      "input_tokens": 234,
      "output_tokens": 39,
      "reasoning_tokens": 3076,
      "undefined_reason": null
    },
    {
      "snapshot_id": "87e2297a-ee74-4ba8-808e-a82867431f6a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00184978,
      "cost_per_success_usd": 0.0006165933333333334,
      "octane_score": 2.9971131702148357,
      "ci_low": 2.797019035706402,
      "ci_high": 3.214889073503091,
      "ci_metric": "octane_score",
      "median_latency_ms": 4007.5829587876797,
      "p95_latency_ms": 4050.9786715731025,
      "input_tokens": 276,
      "output_tokens": 52,
      "reasoning_tokens": 746,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6f27732f-e60f-4efb-8529-fcaef1772065",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00172024,
      "cost_per_success_usd": 0.0005734133333333333,
      "octane_score": 2.07296656280519,
      "ci_low": 1.8642861477856432,
      "ci_high": 2.305296715581791,
      "ci_metric": "octane_score",
      "median_latency_ms": 4058.395417407155,
      "p95_latency_ms": 4583.058778941631,
      "input_tokens": 258,
      "output_tokens": 15,
      "reasoning_tokens": 727,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e8b320a1-ad02-4946-b4a3-352643ab4ee4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00240729,
      "cost_per_success_usd": 0.0008024299999999999,
      "octane_score": 3.11636736745469,
      "ci_low": 2.422559117386105,
      "ci_high": 4.412915218145775,
      "ci_metric": "octane_score",
      "median_latency_ms": 5265.046499669552,
      "p95_latency_ms": 7148.609212785959,
      "input_tokens": 468,
      "output_tokens": 12,
      "reasoning_tokens": 1017,
      "undefined_reason": null
    },
    {
      "snapshot_id": "30ffb383-dd1b-437e-8055-ec095f4717ee",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0035487,
      "cost_per_success_usd": 0.0011829,
      "octane_score": 2.8658381942683233,
      "ci_low": 2.1232619316046595,
      "ci_high": 3.9458029229485194,
      "ci_metric": "octane_score",
      "median_latency_ms": 11153.037499636412,
      "p95_latency_ms": 12183.894987404346,
      "input_tokens": 491,
      "output_tokens": 67,
      "reasoning_tokens": 1412,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6e01a71b-641c-4caa-adcd-472f4d27e5d4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0067494,
      "cost_per_success_usd": 0.0022498,
      "octane_score": 1.5068006044981774,
      "ci_low": 1.2383322962041072,
      "ci_high": 1.7825778122174865,
      "ci_metric": "octane_score",
      "median_latency_ms": 2124.4735829532146,
      "p95_latency_ms": 2912.0238324627276,
      "input_tokens": 723,
      "output_tokens": 64,
      "reasoning_tokens": 322,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bd49f9ed-de2c-40eb-91f5-399117f90541",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0044069999999999995,
      "cost_per_success_usd": 0.0014689999999999998,
      "octane_score": 0.8091672339459951,
      "ci_low": 0.6412008045846143,
      "ci_high": 1.2132301686881362,
      "ci_metric": "octane_score",
      "median_latency_ms": 1713.5724583640695,
      "p95_latency_ms": 1844.551445171237,
      "input_tokens": 492,
      "output_tokens": 3,
      "reasoning_tokens": 250,
      "undefined_reason": null
    },
    {
      "snapshot_id": "11dede52-7e38-4be2-ab34-40a1f2de8a81",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.008478,
      "cost_per_success_usd": 0.002826,
      "octane_score": 0.6539278131634819,
      "ci_low": 0.5047311827956992,
      "ci_high": 0.840381991814461,
      "ci_metric": "octane_score",
      "median_latency_ms": 5777.751541696489,
      "p95_latency_ms": 13173.20829182863,
      "input_tokens": 551,
      "output_tokens": 61,
      "reasoning_tokens": 394,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5ba71eec-fd6c-4681-948c-2a13620fb980",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00801,
      "cost_per_success_usd": 0.00267,
      "octane_score": 5.647565543071161,
      "ci_low": 3.5400398101362733,
      "ci_high": 8.754681647940075,
      "ci_metric": "octane_score",
      "median_latency_ms": 1760.5849578976631,
      "p95_latency_ms": 1892.1333458274603,
      "input_tokens": 468,
      "output_tokens": 228,
      "reasoning_tokens": 270,
      "undefined_reason": null
    },
    {
      "snapshot_id": "26ac26e2-12b0-4327-9571-606850b10a03",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007016399999999999,
      "cost_per_success_usd": 0.0023388,
      "octane_score": 1.0692092811128215,
      "ci_low": 0.9995822890559733,
      "ci_high": 1.1471861471861473,
      "ci_metric": "octane_score",
      "median_latency_ms": 1522.3202910274267,
      "p95_latency_ms": 1722.1102780662477,
      "input_tokens": 702,
      "output_tokens": 0,
      "reasoning_tokens": 408,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fa12d093-42ae-4a11-a825-58528ccf1dff",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 3,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00031800000000000003,
      "cost_per_success_usd": 0.00015900000000000002,
      "octane_score": 15.727463312368972,
      "ci_low": 4.502496878901374,
      "ci_high": 76.38235294117646,
      "ci_metric": "octane_score",
      "median_latency_ms": 563.224290497601,
      "p95_latency_ms": 8349.162466358393,
      "input_tokens": 474,
      "output_tokens": 279,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d1771418-e808-4a7a-b7e5-b619e810daf5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 6.300000000000001e-05,
      "cost_per_success_usd": 2.1000000000000006e-05,
      "octane_score": 56.60317460317459,
      "ci_low": 52.28571428571427,
      "ci_high": 61.19047619047618,
      "ci_metric": "octane_score",
      "median_latency_ms": 661.4632504060864,
      "p95_latency_ms": 775.486350338906,
      "input_tokens": 267,
      "output_tokens": 12,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "643aa839-e122-4bdc-b48e-08942808b438",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0008220000000000001,
      "cost_per_success_usd": 0.00027400000000000005,
      "octane_score": 55.03284671532845,
      "ci_low": 33.75182481751824,
      "ci_high": 85.31021897810217,
      "ci_metric": "octane_score",
      "median_latency_ms": 3569.8713744059205,
      "p95_latency_ms": 3597.6172123104334,
      "input_tokens": 234,
      "output_tokens": 969,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d0218d49-a22f-46c2-8c0b-9706098f621d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00045700000000000005,
      "cost_per_success_usd": 0.00015233333333333334,
      "octane_score": 12.131291028446388,
      "ci_low": 11.308967275818102,
      "ci_high": 13.004059539918806,
      "ci_metric": "octane_score",
      "median_latency_ms": 2431.12241756171,
      "p95_latency_ms": 7892.627704050391,
      "input_tokens": 285,
      "output_tokens": 500,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "aa77e3c3-a9cd-4c18-b777-495a0f1720de",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.000786,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 9114.929041825235,
      "p95_latency_ms": 12053.222903702406,
      "input_tokens": 498,
      "output_tokens": 858,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "de3acc9b-6d0f-478f-9c81-13080cec24ab",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 33.71645789593458,
      "p95_latency_ms": 40.27768429368734,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "c1d1ae92-e178-4114-a10e-1ae3f9b10886",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.174708925187588,
      "p95_latency_ms": 35.856258776038885,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "e3132f65-e920-4c84-824c-4859e7ff9d86",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 32.43458364158869,
      "p95_latency_ms": 34.59274601191282,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "14a3eece-4f35-491c-a875-51de1b34a554",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 27.757207863032818,
      "p95_latency_ms": 31.51380755007267,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "58693f4f-f517-47a9-b88d-fa92ed49a797",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 28.36024947464466,
      "p95_latency_ms": 30.597537849098444,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "2d673e41-1db1-49fc-9ef9-9d2e88ab93cf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.035500000000000004,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 2928.590500727296,
      "p95_latency_ms": 3081.833200342953,
      "input_tokens": 690,
      "output_tokens": 467,
      "reasoning_tokens": 105,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "11a46bc0-35aa-4974-b039-981c21d464eb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.01616,
      "cost_per_success_usd": 0.005386666666666667,
      "octane_score": 2.7993193069306925,
      "ci_low": 1.754838709677419,
      "ci_high": 4.247425802543912,
      "ci_metric": "octane_score",
      "median_latency_ms": 2838.035416789353,
      "p95_latency_ms": 3002.4448675103486,
      "input_tokens": 336,
      "output_tokens": 45,
      "reasoning_tokens": 211,
      "undefined_reason": null
    },
    {
      "snapshot_id": "df9af5a8-b260-4ca1-a1fd-73237753aafe",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.03952,
      "cost_per_success_usd": 0.013173333333333334,
      "octane_score": 0.14028340080971657,
      "ci_low": 0.12741201819487669,
      "ci_high": 0.15470888113764422,
      "ci_metric": "octane_score",
      "median_latency_ms": 2843.914167024195,
      "p95_latency_ms": 2868.0961169302464,
      "input_tokens": 447,
      "output_tokens": 460,
      "reasoning_tokens": 241,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9cef7edd-569a-4277-89aa-e234e674cb43",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00498,
      "cost_per_success_usd": 0.00166,
      "octane_score": 0.7160642570281124,
      "ci_low": 0.6614457831325301,
      "ci_high": 0.7740963855421688,
      "ci_metric": "octane_score",
      "median_latency_ms": 1779.810500331223,
      "p95_latency_ms": 2144.8967375792563,
      "input_tokens": 378,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7b48f2b9-89ec-4dc3-bcba-3a6af1f1f29c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.01779,
      "cost_per_success_usd": 0.00593,
      "octane_score": 0.5716694772344013,
      "ci_low": 0.4792431192660552,
      "ci_high": 0.6388729154686599,
      "ci_metric": "octane_score",
      "median_latency_ms": 2821.426250040531,
      "p95_latency_ms": 3373.671123944223,
      "input_tokens": 759,
      "output_tokens": 18,
      "reasoning_tokens": 186,
      "undefined_reason": null
    },
    {
      "snapshot_id": "70b7273d-7c5c-41c7-aae1-bb7188571996",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00034372,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3004.6389577910304,
      "p95_latency_ms": 3213.707720860839,
      "input_tokens": 300,
      "output_tokens": 1247,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "6542a139-8cfe-4fcd-a708-b08c4afe207e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000433875,
      "cost_per_success_usd": 0.000144625,
      "octane_score": 17.29069432440219,
      "ci_low": 14.371942672426123,
      "ci_high": 20.04708788451889,
      "ci_metric": "octane_score",
      "median_latency_ms": 4175.9244576096535,
      "p95_latency_ms": 4721.618494484574,
      "input_tokens": 519,
      "output_tokens": 1539,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "697db793-0621-486c-83a7-eca28ff1909a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0008498749999999999,
      "cost_per_success_usd": 0.00028329166666666666,
      "octane_score": 11.966465656714222,
      "ci_low": 10.176859151928406,
      "ci_high": 13.527746491735408,
      "ci_metric": "octane_score",
      "median_latency_ms": 5152.280375361443,
      "p95_latency_ms": 5174.262162018567,
      "input_tokens": 519,
      "output_tokens": 3139,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b48ff468-e38c-4f67-a8d9-413e49231c92",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.003845075,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 25276.151250116527,
      "p95_latency_ms": 27041.934336535633,
      "input_tokens": 267,
      "output_tokens": 14722,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "ac96f559-92e0-4861-966d-76cc0fecfcf7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00034651500000000004,
      "cost_per_success_usd": 0.00011550500000000001,
      "octane_score": 10.29104079188491,
      "ci_low": 9.506081987792735,
      "ci_high": 11.125059521232847,
      "ci_metric": "octane_score",
      "median_latency_ms": 3033.9419171214104,
      "p95_latency_ms": 3123.507667146623,
      "input_tokens": 279,
      "output_tokens": 1263,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bf96c8c4-2609-41da-98b3-28cb2b323118",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003804,
      "cost_per_success_usd": 0.001268,
      "octane_score": 2.673501577287066,
      "ci_low": 2.2041139240506338,
      "ci_high": 2.9438261791202964,
      "ci_metric": "octane_score",
      "median_latency_ms": 1720.4125002026558,
      "p95_latency_ms": 4411.572137102485,
      "input_tokens": 654,
      "output_tokens": 78,
      "reasoning_tokens": 338,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ea18a490-4215-43ce-baec-962587f6d098",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002586,
      "cost_per_success_usd": 0.000862,
      "octane_score": 2.1438515081206493,
      "ci_low": 1.9711111111111108,
      "ci_high": 2.345177664974619,
      "ci_metric": "octane_score",
      "median_latency_ms": 2847.6716242730618,
      "p95_latency_ms": 3007.440337073058,
      "input_tokens": 444,
      "output_tokens": 38,
      "reasoning_tokens": 245,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3a6a14d6-5c7b-46d7-b011-8d6591191b00",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.007103999999999999,
      "cost_per_success_usd": 0.002368,
      "octane_score": 6.367820945945946,
      "ci_low": 3.9186440677966106,
      "ci_high": 9.837962962962962,
      "ci_metric": "octane_score",
      "median_latency_ms": 4816.457124426961,
      "p95_latency_ms": 7355.093499738723,
      "input_tokens": 402,
      "output_tokens": 216,
      "reasoning_tokens": 834,
      "undefined_reason": null
    },
    {
      "snapshot_id": "605ecce3-72a7-45b5-be1a-2cf131a19b7a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0017100000000000001,
      "cost_per_success_usd": 0.0005700000000000001,
      "octane_score": 2.085380116959064,
      "ci_low": 1.8515057113187954,
      "ci_high": 2.351503759398496,
      "ci_metric": "octane_score",
      "median_latency_ms": 3019.685042090714,
      "p95_latency_ms": 3302.143590990454,
      "input_tokens": 432,
      "output_tokens": 20,
      "reasoning_tokens": 121,
      "undefined_reason": null
    },
    {
      "snapshot_id": "dbeb21b4-04b6-45e0-9f05-36456f733c79",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00375,
      "cost_per_success_usd": 0.00125,
      "octane_score": 2.0005333333333333,
      "ci_low": 1.7731071779744343,
      "ci_high": 2.294189602446483,
      "ci_metric": "octane_score",
      "median_latency_ms": 1861.8887094780803,
      "p95_latency_ms": 3478.7335715256627,
      "input_tokens": 654,
      "output_tokens": 4,
      "reasoning_tokens": 403,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d4fce36a-02d6-4db2-93d0-98f47d896921",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0015030500000000001,
      "cost_per_success_usd": 0.0005010166666666667,
      "octane_score": 6.766241974651541,
      "ci_low": 5.647850696037938,
      "ci_high": 7.865512697225486,
      "ci_metric": "octane_score",
      "median_latency_ms": 2797.4227080121636,
      "p95_latency_ms": 5346.030821464955,
      "input_tokens": 645,
      "output_tokens": 71,
      "reasoning_tokens": 308,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ea30acff-412e-41fc-ab4b-39a835fd2aad",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00309025,
      "cost_per_success_usd": 0.0010300833333333334,
      "octane_score": 14.638621470754792,
      "ci_low": 8.977914408219398,
      "ci_high": 22.69233880753984,
      "ci_metric": "octane_score",
      "median_latency_ms": 6007.119292393327,
      "p95_latency_ms": 13445.09762926027,
      "input_tokens": 389,
      "output_tokens": 346,
      "reasoning_tokens": 565,
      "undefined_reason": null
    },
    {
      "snapshot_id": "116ca49d-d503-4eb4-a1ec-976afba8b132",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0011165,
      "cost_per_success_usd": 0.0003721666666666667,
      "octane_score": 4.96551724137931,
      "ci_low": 4.620995050794478,
      "ci_high": 5.332470174789604,
      "ci_metric": "octane_score",
      "median_latency_ms": 2154.393583536148,
      "p95_latency_ms": 2774.6950704604383,
      "input_tokens": 426,
      "output_tokens": 91,
      "reasoning_tokens": 198,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5ad73d5f-ba6b-49f9-bf73-ac3a53bffc69",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00181485,
      "cost_per_success_usd": 0.00060495,
      "octane_score": 4.133674959363033,
      "ci_low": 3.8691162665951455,
      "ci_high": 4.433424249938011,
      "ci_metric": "octane_score",
      "median_latency_ms": 2968.913790769875,
      "p95_latency_ms": 3835.4264036752284,
      "input_tokens": 641,
      "output_tokens": 156,
      "reasoning_tokens": 321,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fb5d00b4-017b-4872-982a-42ab9ce62882",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005022500000000001,
      "cost_per_success_usd": 0.0001674166666666667,
      "octane_score": 7.100049776007962,
      "ci_low": 6.558486809357888,
      "ci_high": 7.675460428073667,
      "ci_metric": "octane_score",
      "median_latency_ms": 1529.0316669270396,
      "p95_latency_ms": 1612.9256915301085,
      "input_tokens": 405,
      "output_tokens": 0,
      "reasoning_tokens": 100,
      "undefined_reason": null
    },
    {
      "snapshot_id": "65c6157e-9a95-4ec4-9d9e-60a4ca66e77d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003282,
      "cost_per_success_usd": 0.0010940000000000001,
      "octane_score": 2.285801340645947,
      "ci_low": 2.101689976689977,
      "ci_high": 2.4875478927203063,
      "ci_metric": "octane_score",
      "median_latency_ms": 3761.747917160392,
      "p95_latency_ms": 3862.190241832286,
      "input_tokens": 633,
      "output_tokens": 21,
      "reasoning_tokens": 315,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f3d0107d-5bda-4274-b800-97405de64b93",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001512,
      "cost_per_success_usd": 0.000504,
      "octane_score": 2.3584656084656084,
      "ci_low": 2.1872509960159365,
      "ci_high": 2.549603174603175,
      "ci_metric": "octane_score",
      "median_latency_ms": 1832.2858745232224,
      "p95_latency_ms": 2270.1617499813437,
      "input_tokens": 393,
      "output_tokens": 24,
      "reasoning_tokens": 97,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4db21d97-ac8a-4336-ad80-058ed6626c37",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002538,
      "cost_per_success_usd": 0.000846,
      "octane_score": 2.1843971631205674,
      "ci_low": 2.006787330316742,
      "ci_high": 2.4062499999999996,
      "ci_metric": "octane_score",
      "median_latency_ms": 2871.951624751091,
      "p95_latency_ms": 3333.1824622116983,
      "input_tokens": 414,
      "output_tokens": 57,
      "reasoning_tokens": 228,
      "undefined_reason": null
    },
    {
      "snapshot_id": "936541db-8c5c-48fe-9c7e-8927ecfc3c9d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.008622,
      "cost_per_success_usd": 0.002874,
      "octane_score": 5.24669450243563,
      "ci_low": 3.3099498926270576,
      "ci_high": 7.999657768651609,
      "ci_metric": "octane_score",
      "median_latency_ms": 7817.1730833128095,
      "p95_latency_ms": 8253.153445851058,
      "input_tokens": 381,
      "output_tokens": 54,
      "reasoning_tokens": 1256,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f202abfc-b3e3-4cd1-a566-2f3efc834921",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003726,
      "cost_per_success_usd": 0.001242,
      "octane_score": 2.729468599033816,
      "ci_low": 2.2857850421000494,
      "ci_high": 3.2430603057004777,
      "ci_metric": "octane_score",
      "median_latency_ms": 3532.551917247474,
      "p95_latency_ms": 3644.452029466629,
      "input_tokens": 633,
      "output_tokens": 24,
      "reasoning_tokens": 386,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c3cd39a5-04b4-4ab8-b56e-d61de2618c38",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005925,
      "cost_per_success_usd": 0.0001975,
      "octane_score": 6.018565400843881,
      "ci_low": 4.9292487235594455,
      "ci_high": 8.438787878787878,
      "ci_metric": "octane_score",
      "median_latency_ms": 1363.4298322722316,
      "p95_latency_ms": 1531.585495825857,
      "input_tokens": 357,
      "output_tokens": 12,
      "reasoning_tokens": 126,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bf0c08d0-5fe1-4a6b-b249-a138d70407ee",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016020000000000001,
      "cost_per_success_usd": 0.0005340000000000001,
      "octane_score": 3.46067415730337,
      "ci_low": 2.3042394014962593,
      "ci_high": 4.853535353535353,
      "ci_metric": "octane_score",
      "median_latency_ms": 2214.2891250550747,
      "p95_latency_ms": 3770.9845497272904,
      "input_tokens": 366,
      "output_tokens": 45,
      "reasoning_tokens": 428,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c07585ea-4649-4071-8550-04f768cc60d5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0346515,
      "cost_per_success_usd": 0.0115505,
      "octane_score": 1.3054846110557983,
      "ci_low": 0.668897878408661,
      "ci_high": 3.175529114457197,
      "ci_metric": "octane_score",
      "median_latency_ms": 58461.9215419516,
      "p95_latency_ms": 77706.0283667408,
      "input_tokens": 339,
      "output_tokens": 42,
      "reasoning_tokens": 11452,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0f6836ad-4e89-495a-afe2-15dff02375ed",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009627,
      "cost_per_success_usd": 0.003209,
      "octane_score": 1.0564038641321283,
      "ci_low": 0.6690349319123741,
      "ci_high": 1.6387652122291483,
      "ci_metric": "octane_score",
      "median_latency_ms": 10934.03562437743,
      "p95_latency_ms": 22320.368374977257,
      "input_tokens": 594,
      "output_tokens": 12,
      "reasoning_tokens": 3098,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8fb86b8a-e437-40d7-b6ad-3194f1ab57a1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0025725,
      "cost_per_success_usd": 0.0008575000000000001,
      "octane_score": 2.91622934888241,
      "ci_low": 1.8504604051565376,
      "ci_high": 4.612209802235598,
      "ci_metric": "octane_score",
      "median_latency_ms": 3315.4466254636645,
      "p95_latency_ms": 6345.654487889259,
      "input_tokens": 585,
      "output_tokens": 6,
      "reasoning_tokens": 754,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5e1e4aff-55a6-44ef-a6e9-4878483a8941",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-defective-return-exception",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0011875,
      "cost_per_success_usd": 0.0003958333333333333,
      "octane_score": 6.317473684210526,
      "ci_low": 5.735984095427436,
      "ci_high": 7.144761904761905,
      "ci_metric": "octane_score",
      "median_latency_ms": 3599.64391682297,
      "p95_latency_ms": 4144.832829199731,
      "input_tokens": 630,
      "output_tokens": 6,
      "reasoning_tokens": 343,
      "undefined_reason": null
    },
    {
      "snapshot_id": "79c3113d-93fb-428c-a1b0-bb5576db5dd0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-escalation-decision",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009781999999999999,
      "cost_per_success_usd": 0.0032606666666666665,
      "octane_score": 1.0396646902473932,
      "ci_low": 0.59438590379982,
      "ci_high": 2.541229385307346,
      "ci_metric": "octane_score",
      "median_latency_ms": 14116.682375781238,
      "p95_latency_ms": 32997.007775027305,
      "input_tokens": 639,
      "output_tokens": 12,
      "reasoning_tokens": 3773,
      "undefined_reason": null
    },
    {
      "snapshot_id": "68f9b844-35b1-4e3a-9496-30f65a5c9795",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-prorated-refund",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0236195,
      "cost_per_success_usd": 0.007873166666666667,
      "octane_score": 1.9152395266622915,
      "ci_low": 1.0091802640542382,
      "ci_high": 5.058716926422617,
      "ci_metric": "octane_score",
      "median_latency_ms": 50799.52466674149,
      "p95_latency_ms": 54116.23781649396,
      "input_tokens": 384,
      "output_tokens": 42,
      "reasoning_tokens": 9329,
      "undefined_reason": null
    },
    {
      "snapshot_id": "11fe4559-a3c3-47d4-9059-d998977a7acf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-refund-eligibility",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001113,
      "cost_per_success_usd": 0.000371,
      "octane_score": 4.981132075471697,
      "ci_low": 4.625814863102999,
      "ci_high": 5.361227336122733,
      "ci_metric": "octane_score",
      "median_latency_ms": 4609.73387490958,
      "p95_latency_ms": 4757.969387341291,
      "input_tokens": 411,
      "output_tokens": 45,
      "reasoning_tokens": 318,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c04009b2-dae3-4c28-9a06-03a56e9d37b5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "customer_service",
      "task_id": "cs-ticket-intent-classification",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000591,
      "cost_per_success_usd": 0.00019700000000000002,
      "octane_score": 6.033840947546531,
      "ci_low": 4.2404181184668985,
      "ci_high": 8.230263157894736,
      "ci_metric": "octane_score",
      "median_latency_ms": 2075.100624933839,
      "p95_latency_ms": 2587.25184937939,
      "input_tokens": 402,
      "output_tokens": 12,
      "reasoning_tokens": 144,
      "undefined_reason": null
    },
    {
      "snapshot_id": "84cd4868-0707-45e8-a3dd-dfd7ba2fdad3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.000762,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 588.2025407627225,
      "p95_latency_ms": 625.814592000097,
      "input_tokens": 612,
      "output_tokens": 30,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "89742705-71aa-47d9-8d8a-038995f789aa",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.001209,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 646.2804572656751,
      "p95_latency_ms": 649.9920201487839,
      "input_tokens": 594,
      "output_tokens": 123,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "b95612da-149f-4362-81e2-2f0badb75cb4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000675,
      "cost_per_success_usd": 0.00022500000000000002,
      "octane_score": 18.90222222222222,
      "ci_low": 17.977777777777774,
      "ci_high": 19.36444444444444,
      "ci_metric": "octane_score",
      "median_latency_ms": 688.6766655370593,
      "p95_latency_ms": 710.1972790434957,
      "input_tokens": 585,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8f3463e4-4cf7-47cd-9acf-f228c3ffbcdd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.000876,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 576.4287495985627,
      "p95_latency_ms": 618.9523253589869,
      "input_tokens": 441,
      "output_tokens": 87,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "d65750ab-c414-42d4-84fb-1b0b89bb8e17",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0009540000000000001,
      "cost_per_success_usd": 0.00031800000000000003,
      "octane_score": 7.504192872117399,
      "ci_low": 6.2924528301886795,
      "ci_high": 9.71383647798742,
      "ci_metric": "octane_score",
      "median_latency_ms": 611.1413324251771,
      "p95_latency_ms": 667.6976708695292,
      "input_tokens": 369,
      "output_tokens": 117,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9353f236-6c4b-4d22-9f31-0d8f61fdbdd6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0046700000000000005,
      "cost_per_success_usd": 0.0023350000000000003,
      "octane_score": 1.0219842969307635,
      "ci_low": 0.4287429378531073,
      "ci_high": 1.7850649350649344,
      "ci_metric": "octane_score",
      "median_latency_ms": 1996.1476251482964,
      "p95_latency_ms": 2006.378937419504,
      "input_tokens": 369,
      "output_tokens": 113,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "46fee6ec-bb44-44c8-9100-affb7a33bb2a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00381,
      "cost_per_success_usd": 0.00127,
      "octane_score": 9.09711286089239,
      "ci_low": 7.133858267716534,
      "ci_high": 10.078740157480317,
      "ci_metric": "octane_score",
      "median_latency_ms": 1614.7413328289986,
      "p95_latency_ms": 1660.869595501572,
      "input_tokens": 612,
      "output_tokens": 30,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e884de9b-7d89-437f-955c-d9378a0000d9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0033749999999999995,
      "cost_per_success_usd": 0.001125,
      "octane_score": 3.7804444444444445,
      "ci_low": 3.595555555555556,
      "ci_high": 3.872888888888889,
      "ci_metric": "octane_score",
      "median_latency_ms": 1790.9887926653028,
      "p95_latency_ms": 1887.2559798881412,
      "input_tokens": 585,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d5a3c086-186f-425d-894c-1503f9d734c8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.007095000000000001,
      "cost_per_success_usd": 0.0023650000000000003,
      "octane_score": null,
      "ci_low": 0.00219,
      "ci_high": 0.002715,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 2458.8842503726482,
      "p95_latency_ms": 2713.0450759083033,
      "input_tokens": 594,
      "output_tokens": 165,
      "reasoning_tokens": 0,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "940dfa80-557d-4554-b709-0e4ca5af2c60",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.02303,
      "cost_per_success_usd": 0.007676666666666667,
      "octane_score": 3.143942683456361,
      "ci_low": 1.8151188782767735,
      "ci_high": 5.439649272882807,
      "ci_metric": "octane_score",
      "median_latency_ms": 5772.119750268757,
      "p95_latency_ms": 5948.982312716544,
      "input_tokens": 441,
      "output_tokens": 833,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "44bc21d4-2de5-426d-9f01-60d7389a58cc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.005076,
      "cost_per_success_usd": 0.001692,
      "octane_score": 6.828211189913319,
      "ci_low": 5.354609929078014,
      "ci_high": 8.465608465608467,
      "ci_metric": "octane_score",
      "median_latency_ms": 2302.7822077274323,
      "p95_latency_ms": 2931.571708526462,
      "input_tokens": 612,
      "output_tokens": 216,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bf510e4a-d78e-4f28-851e-6a21d9073985",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.014748,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 5961.780250072479,
      "p95_latency_ms": 6282.738099806011,
      "input_tokens": 441,
      "output_tokens": 895,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "c2ca79f6-18ca-4e24-a96f-8fd97f84032c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002025,
      "cost_per_success_usd": 0.0006749999999999999,
      "octane_score": 6.300740740740741,
      "ci_low": 5.992592592592593,
      "ci_high": 6.4548148148148154,
      "ci_metric": "octane_score",
      "median_latency_ms": 1131.8331249058247,
      "p95_latency_ms": 1556.4053123816846,
      "input_tokens": 585,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b0b7e86b-6f10-47fb-9032-f2a55f182d98",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004017,
      "cost_per_success_usd": 0.001339,
      "octane_score": null,
      "ci_low": 0.001284,
      "ci_high": 0.0014490000000000002,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 1821.3459169492126,
      "p95_latency_ms": 2475.5238546058536,
      "input_tokens": 594,
      "output_tokens": 149,
      "reasoning_tokens": 0,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "fac1d26d-dce9-448b-8bf9-e35585fe3509",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.002562,
      "cost_per_success_usd": 0.002562,
      "octane_score": 0.9314337756960707,
      "ci_low": 0.7810304449648713,
      "ci_high": 2.8613109512390085,
      "ci_metric": "octane_score",
      "median_latency_ms": 1550.054875202477,
      "p95_latency_ms": 2352.9104497283697,
      "input_tokens": 369,
      "output_tokens": 97,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "63c3dac0-1093-4164-8782-50f0cafed64d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0003656600000000001,
      "cost_per_success_usd": 0.00012188666666666669,
      "octane_score": 19.578296778428044,
      "ci_low": 9.91249757611014,
      "ci_high": 61.453055141579725,
      "ci_metric": "octane_score",
      "median_latency_ms": 1983.9939577504992,
      "p95_latency_ms": 16189.054821152238,
      "input_tokens": 357,
      "output_tokens": 718,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "520cfa33-d511-42d2-b1e6-ab8fa519a3d9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.00013408,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 868.8944578170776,
      "p95_latency_ms": 949.2424822412431,
      "input_tokens": 549,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "f3910c82-ea86-4bef-ac73-a47e53e42da8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00014148,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 1295.9378343075514,
      "p95_latency_ms": 1325.1945085823536,
      "input_tokens": 417,
      "output_tokens": 87,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "97aa940b-6462-484c-9d22-fa35f938bd9e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00013914,
      "cost_per_success_usd": 4.638e-05,
      "octane_score": 91.69900819318671,
      "ci_low": 87.21431651573954,
      "ci_high": 93.94135403191031,
      "ci_metric": "octane_score",
      "median_latency_ms": 863.1053324788809,
      "p95_latency_ms": 894.5036333054304,
      "input_tokens": 522,
      "output_tokens": 9,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "410cdb5f-26d9-4060-9970-039ff8ef4ef6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00015646,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 4321.5915421023965,
      "p95_latency_ms": 7392.357341013848,
      "input_tokens": 543,
      "output_tokens": 84,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "0bfc7724-286c-4c95-9a86-5a3e53e600cc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00058232,
      "cost_per_success_usd": 0.00019410666666666666,
      "octane_score": 59.52053853551313,
      "ci_low": 35.57279799659257,
      "ci_high": 150.16029806775848,
      "ci_metric": "octane_score",
      "median_latency_ms": 17921.87541630119,
      "p95_latency_ms": 31631.463030073788,
      "input_tokens": 628,
      "output_tokens": 527,
      "reasoning_tokens": 2429,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d76de25b-f48f-4cf8-8c0f-b6ce76a4e7b9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00013526,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 1393.9756657928228,
      "p95_latency_ms": 12036.449490953235,
      "input_tokens": 622,
      "output_tokens": 76,
      "reasoning_tokens": 399,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "608dc216-a0fc-46ce-b9c3-4cd191e233ac",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 5.946e-05,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 1031.674916855991,
      "p95_latency_ms": 2334.7383411601186,
      "input_tokens": 345,
      "output_tokens": 51,
      "reasoning_tokens": 126,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "c7963a12-9fce-4a1f-abbf-2b4f8b6865ac",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00016879599999999997,
      "cost_per_success_usd": 5.6265333333333325e-05,
      "octane_score": 75.5882840825612,
      "ci_low": 52.77984611566146,
      "ci_high": 108.22984527687296,
      "ci_metric": "octane_score",
      "median_latency_ms": 4550.234083086252,
      "p95_latency_ms": 5223.587233573197,
      "input_tokens": 662,
      "output_tokens": 358,
      "reasoning_tokens": 373,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a0bc3180-1b1e-45f7-a179-0269bd290c45",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005247800000000001,
      "cost_per_success_usd": 0.0001749266666666667,
      "octane_score": 137.97210259537326,
      "ci_low": 69.9972232506479,
      "ci_high": 249.54898833848304,
      "ci_metric": "octane_score",
      "median_latency_ms": 28898.073707707226,
      "p95_latency_ms": 69208.29115817323,
      "input_tokens": 478,
      "output_tokens": 167,
      "reasoning_tokens": 2536,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0f3e23fc-4107-4647-95d6-5211c2f4009f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.012696403,
      "cost_per_success_usd": 0.004232134333333333,
      "octane_score": 2.7299070453261454,
      "ci_low": 1.1897163744996482,
      "ci_high": 11.969079302920273,
      "ci_metric": "octane_score",
      "median_latency_ms": 32878.931084647775,
      "p95_latency_ms": 155480.40440846232,
      "input_tokens": 707,
      "output_tokens": 138,
      "reasoning_tokens": 14229,
      "undefined_reason": null
    },
    {
      "snapshot_id": "322f8eed-7ab4-4540-9a39-4cd4eb74b3f4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0004902449999999999,
      "cost_per_success_usd": 0.00024512249999999997,
      "octane_score": 9.735268420211664,
      "ci_low": 5.708998251683221,
      "ci_high": 13.257776706052567,
      "ci_metric": "octane_score",
      "median_latency_ms": 3036.3294994458556,
      "p95_latency_ms": 3170.6740005873144,
      "input_tokens": 503,
      "output_tokens": 70,
      "reasoning_tokens": 242,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5cd2c869-57ea-4886-b3e1-da053ef258ca",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00629793,
      "cost_per_success_usd": 0.00209931,
      "octane_score": 11.496634608514224,
      "ci_low": 5.14683003395034,
      "ci_high": 38.30936693438497,
      "ci_metric": "octane_score",
      "median_latency_ms": 54173.937750048935,
      "p95_latency_ms": 58535.83916211501,
      "input_tokens": 558,
      "output_tokens": 64,
      "reasoning_tokens": 6896,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7a1990c9-e967-4486-a254-7109694e3ea7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.00411249,
      "cost_per_success_usd": 0.00411249,
      "octane_score": null,
      "ci_low": 0.00087087,
      "ci_high": 0.005469689999999999,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 23798.109457828104,
      "p95_latency_ms": 42443.76222044229,
      "input_tokens": 780,
      "output_tokens": 69,
      "reasoning_tokens": 4268,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "86213631-fd85-4b3e-954c-55886a75ff4d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0012066899999999998,
      "cost_per_success_usd": 0.00040222999999999995,
      "octane_score": 10.573552445118466,
      "ci_low": 7.608569256227917,
      "ci_high": 17.841261850826413,
      "ci_metric": "octane_score",
      "median_latency_ms": 6351.287750527263,
      "p95_latency_ms": 7440.653413534164,
      "input_tokens": 662,
      "output_tokens": 4,
      "reasoning_tokens": 1052,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b58b7f2d-7617-4426-add7-6ffcbec4a34e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00038580000000000005,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 732.9921666532755,
      "p95_latency_ms": 973.2763030566275,
      "input_tokens": 411,
      "output_tokens": 105,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "ac69874c-f251-45d3-9bf0-4b3c667daa13",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.00025049999999999996,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 639.6616250276566,
      "p95_latency_ms": 663.3386373519897,
      "input_tokens": 585,
      "output_tokens": 30,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "240e6a5a-c15e-424a-949d-70c2c5e62eda",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00046380000000000005,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 726.1129580438137,
      "p95_latency_ms": 831.493170838803,
      "input_tokens": 546,
      "output_tokens": 120,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "a26693db-8720-4a95-be3d-5e0e5cd9a238",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0003756,
      "cost_per_success_usd": 0.0001252,
      "octane_score": 19.060170394036206,
      "ci_low": 15.982428115015974,
      "ci_high": 24.672523961661337,
      "ci_metric": "octane_score",
      "median_latency_ms": 730.0700005143881,
      "p95_latency_ms": 891.395787615329,
      "input_tokens": 327,
      "output_tokens": 111,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "26230035-3774-4638-b8f7-009c078ddbb0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00021150000000000002,
      "cost_per_success_usd": 7.05e-05,
      "octane_score": 60.32624113475177,
      "ci_low": 57.37588652482269,
      "ci_high": 61.8014184397163,
      "ci_metric": "octane_score",
      "median_latency_ms": 632.2729578241706,
      "p95_latency_ms": 644.1737700253725,
      "input_tokens": 555,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "54f43fb2-f045-42f4-9d65-6a01b21917ab",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0195945,
      "cost_per_success_usd": 0.0065315,
      "octane_score": 3.6951695628875445,
      "ci_low": 2.086913857152871,
      "ci_high": 6.148869903440265,
      "ci_metric": "octane_score",
      "median_latency_ms": 4854.932040907443,
      "p95_latency_ms": 5615.871916245669,
      "input_tokens": 413,
      "output_tokens": 105,
      "reasoning_tokens": 2425,
      "undefined_reason": null
    },
    {
      "snapshot_id": "071aa852-58af-40d8-900a-22c65b820c76",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.013203,
      "cost_per_success_usd": 0.0044009999999999995,
      "octane_score": 0.9663712792547149,
      "ci_low": 0.8920662223177813,
      "ci_high": 1.0496265960009634,
      "ci_metric": "octane_score",
      "median_latency_ms": 3326.9319171085954,
      "p95_latency_ms": 3454.0578913874924,
      "input_tokens": 557,
      "output_tokens": 18,
      "reasoning_tokens": 1631,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2db0955e-e8d4-4fe7-ac1e-a6f520d215b5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0216555,
      "cost_per_success_usd": 0.0072185,
      "octane_score": 1.600517189628501,
      "ci_low": 1.2871146469668986,
      "ci_high": 1.8236215985183077,
      "ci_metric": "octane_score",
      "median_latency_ms": 4769.888833165169,
      "p95_latency_ms": 5288.391657639294,
      "input_tokens": 587,
      "output_tokens": 30,
      "reasoning_tokens": 2740,
      "undefined_reason": null
    },
    {
      "snapshot_id": "703cbdf0-6de7-42b8-9936-694fcf95c9f0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0117255,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3356.960207223892,
      "p95_latency_ms": 3646.6467704623938,
      "input_tokens": 547,
      "output_tokens": 111,
      "reasoning_tokens": 1343,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "81a73d6f-a735-4a6d-ac18-1f77a3d909d4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.015307499999999998,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3554.0985418483615,
      "p95_latency_ms": 4413.514955062419,
      "input_tokens": 330,
      "output_tokens": 114,
      "reasoning_tokens": 1861,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "f44e9435-4293-4a51-b95d-8338328e3894",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002259375,
      "cost_per_success_usd": 0.000753125,
      "octane_score": 5.647136929460581,
      "ci_low": 5.370954356846473,
      "ci_high": 5.912807463952502,
      "ci_metric": "octane_score",
      "median_latency_ms": 4283.49429089576,
      "p95_latency_ms": 4469.614928588271,
      "input_tokens": 555,
      "output_tokens": 18,
      "reasoning_tokens": 1076,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b9b6e4e3-d28a-4ebe-8785-1db6320668c9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0027045,
      "cost_per_success_usd": 0.00135225,
      "octane_score": 1.7647131324335983,
      "ci_low": 0.7261093170673365,
      "ci_high": 3.1634062140391253,
      "ci_metric": "octane_score",
      "median_latency_ms": 3895.449082367122,
      "p95_latency_ms": 4222.877896297723,
      "input_tokens": 327,
      "output_tokens": 112,
      "reasoning_tokens": 1265,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9b65c89e-09ce-4557-988b-0889d3b97a98",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002887875,
      "cost_per_success_usd": 0.0009626249999999999,
      "octane_score": 25.072068562524347,
      "ci_low": 14.446036294173831,
      "ci_high": 42.54723372224314,
      "ci_metric": "octane_score",
      "median_latency_ms": 4254.869583062828,
      "p95_latency_ms": 6221.905384119599,
      "input_tokens": 411,
      "output_tokens": 99,
      "reasoning_tokens": 1359,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bffb6e66-981c-443c-9b95-34137a64c407",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0014759999999999999,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 2531.6067915409803,
      "p95_latency_ms": 2829.0693167597055,
      "input_tokens": 546,
      "output_tokens": 114,
      "reasoning_tokens": 564,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "705d3df8-3497-4a99-914d-b90e42d974f8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004835625,
      "cost_per_success_usd": 0.0016118749999999998,
      "octane_score": 7.167636034638751,
      "ci_low": 5.428879099363683,
      "ci_high": 9.624060150375943,
      "ci_metric": "octane_score",
      "median_latency_ms": 5366.891874931753,
      "p95_latency_ms": 7091.704900097102,
      "input_tokens": 585,
      "output_tokens": 30,
      "reasoning_tokens": 2432,
      "undefined_reason": null
    },
    {
      "snapshot_id": "51fff0b5-c9e4-46f4-8f12-50605394b57e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.0007616,
      "cost_per_success_usd": 0.0007616,
      "octane_score": null,
      "ci_low": 0.0002904,
      "ci_high": 0.0007935999999999999,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 7082.8834576532245,
      "p95_latency_ms": 15826.779346540568,
      "input_tokens": 540,
      "output_tokens": 200,
      "reasoning_tokens": 1623,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "25a5292e-4dea-49cd-9129-3d437c008019",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0005373600000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 29319.505041465163,
      "p95_latency_ms": 32008.728817477822,
      "input_tokens": 336,
      "output_tokens": 106,
      "reasoning_tokens": 1203,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "bc951b7c-7462-4bb4-9744-ecd696e2ca66",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00309356,
      "cost_per_success_usd": 0.00154678,
      "octane_score": 7.469280268256207,
      "ci_low": 3.772074934000888,
      "ci_high": 15.849933234557065,
      "ci_metric": "octane_score",
      "median_latency_ms": 39407.995582558215,
      "p95_latency_ms": 112416.10452057792,
      "input_tokens": 546,
      "output_tokens": 1288,
      "reasoning_tokens": 6396,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c73a9912-3365-477a-ace1-ebd6a58e0f6e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00097784,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 8315.761833451688,
      "p95_latency_ms": 8317.940582614392,
      "input_tokens": 384,
      "output_tokens": 423,
      "reasoning_tokens": 1980,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "9fb95ce0-3d15-4f41-937f-e429398adcfd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": 1,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.00080824,
      "cost_per_success_usd": 0.00080824,
      "octane_score": 5.262050875977432,
      "ci_low": 4.875211505922167,
      "ci_high": 19.719028189910976,
      "ci_metric": "octane_score",
      "median_latency_ms": 11379.366124980152,
      "p95_latency_ms": 27123.704975005236,
      "input_tokens": 504,
      "output_tokens": 55,
      "reasoning_tokens": 1890,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1a203cd7-2b88-4141-9493-d72803129e0d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00818412,
      "cost_per_success_usd": 0.0027280399999999997,
      "octane_score": 8.847011040893829,
      "ci_low": 5.029941353404114,
      "ci_high": 15.256963938674154,
      "ci_metric": "octane_score",
      "median_latency_ms": 9689.710333943367,
      "p95_latency_ms": 14847.047608718276,
      "input_tokens": 405,
      "output_tokens": 57,
      "reasoning_tokens": 4143,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d42fb7d7-27a9-4047-8c96-3d3203c11484",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": 1,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.06583104,
      "cost_per_success_usd": 0.06583104,
      "octane_score": 0.1754997844988221,
      "ci_low": 0.15655784984704724,
      "ci_high": 4.298498873907394,
      "ci_metric": "octane_score",
      "median_latency_ms": 105550.95879081637,
      "p95_latency_ms": 157853.7181668915,
      "input_tokens": 560,
      "output_tokens": 9,
      "reasoning_tokens": 34103,
      "undefined_reason": null
    },
    {
      "snapshot_id": "567b3fd3-25ec-48d9-a87b-5727ec16b908",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.02085264,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 48859.64350029826,
      "p95_latency_ms": 139040.48304948956,
      "input_tokens": 540,
      "output_tokens": 65,
      "reasoning_tokens": 10667,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "216a713b-ff78-4558-98fd-6670f22fe3d1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.00629064,
      "cost_per_success_usd": 0.00629064,
      "octane_score": 0.3793466695492562,
      "ci_low": 0.28143862576831774,
      "ci_high": 3.2520214409012453,
      "ci_metric": "octane_score",
      "median_latency_ms": 18308.998583815992,
      "p95_latency_ms": 23988.66855818778,
      "input_tokens": 350,
      "output_tokens": 106,
      "reasoning_tokens": 3061,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f6f9c2ed-431e-48c1-a588-e4ec3ad33562",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00417732,
      "cost_per_success_usd": 0.00139244,
      "octane_score": 3.0543506362931256,
      "ci_low": 2.7465643728042983,
      "ci_high": 3.3488593740392303,
      "ci_metric": "octane_score",
      "median_latency_ms": 10453.864499926567,
      "p95_latency_ms": 10548.766425997019,
      "input_tokens": 511,
      "output_tokens": 8,
      "reasoning_tokens": 2008,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6390e7fd-3ee8-4e9b-ac25-50b4b15b6985",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0070249452,
      "cost_per_success_usd": 0.0023416484,
      "octane_score": 4.9338463166943995,
      "ci_low": 3.6390651811183456,
      "ci_high": 6.9649262710508175,
      "ci_metric": "octane_score",
      "median_latency_ms": 7762.334582395852,
      "p95_latency_ms": 162645.6034461036,
      "input_tokens": 567,
      "output_tokens": 77,
      "reasoning_tokens": 2142,
      "undefined_reason": null
    },
    {
      "snapshot_id": "02593c5b-52ce-4508-ad5e-604bab1aec36",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0088395348,
      "cost_per_success_usd": 0.0029465116,
      "octane_score": 8.191041908676008,
      "ci_low": 4.631350475292849,
      "ci_high": 14.356502368198077,
      "ci_metric": "octane_score",
      "median_latency_ms": 8674.554250203073,
      "p95_latency_ms": 9185.151963494718,
      "input_tokens": 405,
      "output_tokens": 334,
      "reasoning_tokens": 2465,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c732dbe1-42f4-430f-a7c1-758ff308f4ea",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00262821,
      "cost_per_success_usd": 0.0008760700000000001,
      "octane_score": 2.723907145928217,
      "ci_low": 2.2840640588080863,
      "ci_high": 3.525973951853162,
      "ci_metric": "octane_score",
      "median_latency_ms": 3602.343915961683,
      "p95_latency_ms": 4522.1928167156875,
      "input_tokens": 357,
      "output_tokens": 81,
      "reasoning_tokens": 750,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8b67c62f-1ada-4ed3-b57f-24b2f70861a8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0019984884000000002,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 2530.3030824288726,
      "p95_latency_ms": 2592.9532455280423,
      "input_tokens": 561,
      "output_tokens": 91,
      "reasoning_tokens": 514,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "4db58ffe-6c4e-4de1-9295-faf1b0b2a06f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0020736155999999998,
      "cost_per_success_usd": 0.0006912051999999999,
      "octane_score": 6.153020839542296,
      "ci_low": 4.684995983659686,
      "ci_high": 8.144944389161727,
      "ci_metric": "octane_score",
      "median_latency_ms": 2560.8430840075016,
      "p95_latency_ms": 3219.473770260811,
      "input_tokens": 525,
      "output_tokens": 12,
      "reasoning_tokens": 618,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d3ab91e3-c7ed-4bd3-97fd-5853b62283ff",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00370116,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 5683.889999985695,
      "p95_latency_ms": 5924.312474671751,
      "input_tokens": 561,
      "output_tokens": 87,
      "reasoning_tokens": 642,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "92bd9a10-b60d-4be9-9b40-51e681352bab",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00307436,
      "cost_per_success_usd": 0.0010247866666666667,
      "octane_score": 4.150132060005984,
      "ci_low": 3.557804918855613,
      "ci_high": 4.6376719036062495,
      "ci_metric": "octane_score",
      "median_latency_ms": 3633.997540920973,
      "p95_latency_ms": 4024.2354036308825,
      "input_tokens": 525,
      "output_tokens": 12,
      "reasoning_tokens": 586,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8dcefb44-eefd-4826-a160-b401804fa578",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.009460360000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 12950.915208086371,
      "p95_latency_ms": 13551.080170925707,
      "input_tokens": 567,
      "output_tokens": 270,
      "reasoning_tokens": 1766,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "187b0bd3-a99a-450f-9880-74c04e7eb2e9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.009990360000000002,
      "cost_per_success_usd": 0.0033301200000000006,
      "octane_score": 7.247486577060283,
      "ci_low": 3.724940752870816,
      "ci_high": 12.730171885697812,
      "ci_metric": "octane_score",
      "median_latency_ms": 13643.496666103601,
      "p95_latency_ms": 14262.806953582913,
      "input_tokens": 405,
      "output_tokens": 60,
      "reasoning_tokens": 2148,
      "undefined_reason": null
    },
    {
      "snapshot_id": "35c11ec0-fa04-452f-aed8-dcf92c13007d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0019290800000000003,
      "cost_per_success_usd": 0.0006430266666666667,
      "octane_score": 3.7110954444605713,
      "ci_low": 2.4061663272995464,
      "ci_high": 6.049203415192323,
      "ci_metric": "octane_score",
      "median_latency_ms": 2766.936500556767,
      "p95_latency_ms": 3839.8705253377557,
      "input_tokens": 357,
      "output_tokens": 81,
      "reasoning_tokens": 277,
      "undefined_reason": null
    },
    {
      "snapshot_id": "52877cf5-7ec5-4d57-8d58-4e36506f5a04",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00060975,
      "cost_per_success_usd": 0.000304875,
      "octane_score": 7.827251605849391,
      "ci_low": 3.2945326278659612,
      "ci_high": 13.625774473358117,
      "ci_metric": "octane_score",
      "median_latency_ms": 787.2159164398909,
      "p95_latency_ms": 828.4231662750244,
      "input_tokens": 339,
      "output_tokens": 79,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bd8a32c0-6934-4522-a4ee-cb0d029e9965",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.000576,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 882.8403344377875,
      "p95_latency_ms": 976.1091338470578,
      "input_tokens": 552,
      "output_tokens": 36,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "f1d84faf-dcd1-4e6a-b9ff-1251a076b2e3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": 1,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.00054225,
      "cost_per_success_usd": 0.00054225,
      "octane_score": 7.843245735361918,
      "ci_low": 7.6514522821576785,
      "ci_high": 23.529737206085752,
      "ci_metric": "octane_score",
      "median_latency_ms": 672.4044596776366,
      "p95_latency_ms": 694.4735078141093,
      "input_tokens": 507,
      "output_tokens": 36,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "15a8337e-b528-45a3-be4d-48c6e6587ad3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.000909,
      "cost_per_success_usd": 0.000909,
      "octane_score": null,
      "ci_low": 0.00033299999999999996,
      "ci_high": 0.000909,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 752.9217079281807,
      "p95_latency_ms": 847.68012072891,
      "input_tokens": 540,
      "output_tokens": 112,
      "reasoning_tokens": 0,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "210f7fe1-63ae-4a78-a5e2-7096270c7923",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0006165000000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 658.1333754584193,
      "p95_latency_ms": 689.0085876919329,
      "input_tokens": 390,
      "output_tokens": 72,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "35554dac-88db-4852-958a-344963878bd2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000552,
      "cost_per_success_usd": 0.000184,
      "octane_score": 62.78985507246378,
      "ci_low": 49.67105263157894,
      "ci_high": 70.48458149779736,
      "ci_metric": "octane_score",
      "median_latency_ms": 3154.6090422198176,
      "p95_latency_ms": 3171.5133292600513,
      "input_tokens": 552,
      "output_tokens": 36,
      "reasoning_tokens": 332,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1cbfd094-f398-42d5-90c1-38f30f6fdfeb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0004332,
      "cost_per_success_usd": 0.0001444,
      "octane_score": 167.13988919667585,
      "ci_low": 99.77882037533514,
      "ci_high": 291.9628099173554,
      "ci_metric": "octane_score",
      "median_latency_ms": 2429.4789591804147,
      "p95_latency_ms": 2738.307320792228,
      "input_tokens": 390,
      "output_tokens": 66,
      "reasoning_tokens": 230,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1367a089-b267-4b7e-acf8-578ac78d7be6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00028859999999999997,
      "cost_per_success_usd": 9.62e-05,
      "octane_score": 44.20997920997921,
      "ci_low": 42.079107505071015,
      "ci_high": 46.05708245243128,
      "ci_metric": "octane_score",
      "median_latency_ms": 1959.287291392684,
      "p95_latency_ms": 2043.645229190588,
      "input_tokens": 507,
      "output_tokens": 24,
      "reasoning_tokens": 132,
      "undefined_reason": null
    },
    {
      "snapshot_id": "aaac35f2-6bc7-4237-bfa0-db398d384501",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001614,
      "cost_per_success_usd": 5.38e-05,
      "octane_score": 44.355638166047086,
      "ci_low": 37.193308550185876,
      "ci_high": 57.416356877323416,
      "ci_metric": "octane_score",
      "median_latency_ms": 975.3516241908073,
      "p95_latency_ms": 982.2031368501484,
      "input_tokens": 339,
      "output_tokens": 78,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b0f584d7-c2be-4513-96b2-538d686e5628",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.00045359999999999997,
      "cost_per_success_usd": 0.00045359999999999997,
      "octane_score": null,
      "ci_low": 0.0001428,
      "ci_high": 0.0004812,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 2324.3298335000873,
      "p95_latency_ms": 2639.9626467376943,
      "input_tokens": 540,
      "output_tokens": 102,
      "reasoning_tokens": 186,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "7035a29d-4a93-48b5-a26d-9bd91039f285",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0031118,
      "cost_per_success_usd": 0.0010372666666666668,
      "octane_score": 11.138247959380424,
      "ci_low": 9.044323173166509,
      "ci_high": 12.78806447315839,
      "ci_metric": "octane_score",
      "median_latency_ms": 4353.59516646713,
      "p95_latency_ms": 4980.70061635226,
      "input_tokens": 7675,
      "output_tokens": 159,
      "reasoning_tokens": 1155,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1a5e7e37-397b-4ede-a2f7-1b91bd568979",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0027572,
      "cost_per_success_usd": 0.0009190666666666667,
      "octane_score": 26.260336573335262,
      "ci_low": 14.77568980348045,
      "ci_high": 43.94574982722877,
      "ci_metric": "octane_score",
      "median_latency_ms": 3980.324709787965,
      "p95_latency_ms": 5337.489420641213,
      "input_tokens": 6880,
      "output_tokens": 249,
      "reasoning_tokens": 902,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9eed7557-28ac-4f4b-a4f6-e4665e39ebeb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0020302000000000002,
      "cost_per_success_usd": 0.0006767333333333334,
      "octane_score": 6.28460250221653,
      "ci_low": 5.977243621318096,
      "ci_high": 6.585550181378475,
      "ci_metric": "octane_score",
      "median_latency_ms": 2950.6675843149424,
      "p95_latency_ms": 3035.4902961291373,
      "input_tokens": 6827,
      "output_tokens": 123,
      "reasoning_tokens": 431,
      "undefined_reason": null
    },
    {
      "snapshot_id": "df7710ef-6c28-4901-a4d4-7439f4302119",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00114684,
      "cost_per_success_usd": 0.00038228000000000003,
      "octane_score": 6.242370339367304,
      "ci_low": 4.062614203922524,
      "ci_high": 10.105874568046465,
      "ci_metric": "octane_score",
      "median_latency_ms": 2063.1121657788754,
      "p95_latency_ms": 2542.0790030620988,
      "input_tokens": 5988,
      "output_tokens": 285,
      "reasoning_tokens": 108,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4608cb20-5d8d-4a13-b1bd-1dd593165700",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0042325999999999996,
      "cost_per_success_usd": 0.0014108666666666665,
      "octane_score": null,
      "ci_low": 0.0013032,
      "ci_high": 0.0015387999999999999,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 8204.952333122492,
      "p95_latency_ms": 8955.404283385724,
      "input_tokens": 7471,
      "output_tokens": 400,
      "reasoning_tokens": 1882,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "99b0323a-ebf5-47e8-aa65-b03725cdc0f1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001578,
      "cost_per_success_usd": 0.000526,
      "octane_score": 4.5367553865652726,
      "ci_low": 3.804182509505704,
      "ci_high": 5.8726235741444865,
      "ci_metric": "octane_score",
      "median_latency_ms": 1877.870874479413,
      "p95_latency_ms": 1893.4878629632294,
      "input_tokens": 339,
      "output_tokens": 90,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d9ec548c-3ac5-4fa1-8ae0-312003549e7b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004334,
      "cost_per_success_usd": 0.0014446666666666668,
      "octane_score": 7.997231195200739,
      "ci_low": 6.615318784766797,
      "ci_high": 9.495548961424333,
      "ci_metric": "octane_score",
      "median_latency_ms": 3860.7324166223407,
      "p95_latency_ms": 4129.926166217774,
      "input_tokens": 552,
      "output_tokens": 36,
      "reasoning_tokens": 287,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ead1c0b9-50c8-49cf-9b1b-2a1d3eb281ac",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00424,
      "cost_per_success_usd": 0.0014133333333333333,
      "octane_score": 17.076650943396224,
      "ci_low": 10.127210884353744,
      "ci_high": 29.85422535211268,
      "ci_metric": "octane_score",
      "median_latency_ms": 4004.1752085089684,
      "p95_latency_ms": 4480.593733489513,
      "input_tokens": 390,
      "output_tokens": 66,
      "reasoning_tokens": 280,
      "undefined_reason": null
    },
    {
      "snapshot_id": "93254f0a-dd02-41a1-a4b0-8923ad11ac60",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001194,
      "cost_per_success_usd": 0.00039799999999999997,
      "octane_score": 10.685929648241206,
      "ci_low": 10.163316582914574,
      "ci_high": 10.947236180904524,
      "ci_metric": "octane_score",
      "median_latency_ms": 1737.2989170253277,
      "p95_latency_ms": 2005.1643420010805,
      "input_tokens": 507,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "700acfac-d0f9-4a01-b5da-37cde7dea57e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0019199999999999998,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 1190.2155419811606,
      "p95_latency_ms": 1414.2826537601647,
      "input_tokens": 540,
      "output_tokens": 84,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "57a44356-42fd-4650-9fe0-d31bd2264f7b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.038015,
      "cost_per_success_usd": 0.012671666666666666,
      "octane_score": 0.3356306721031172,
      "ci_low": 0.3165967188096147,
      "ci_high": 0.360271071579839,
      "ci_metric": "octane_score",
      "median_latency_ms": 3617.423958145082,
      "p95_latency_ms": 3777.854733169079,
      "input_tokens": 6553,
      "output_tokens": 103,
      "reasoning_tokens": 72,
      "undefined_reason": null
    },
    {
      "snapshot_id": "dbcd5951-40f4-437c-9dbf-bacf9183b391",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.06528,
      "cost_per_success_usd": 0.02176,
      "octane_score": 1.109145220588235,
      "ci_low": 0.6715434929704535,
      "ci_high": 1.9316373025516405,
      "ci_metric": "octane_score",
      "median_latency_ms": 6364.836500026286,
      "p95_latency_ms": 7200.066462252289,
      "input_tokens": 6900,
      "output_tokens": 243,
      "reasoning_tokens": 783,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8fb2b083-549d-4448-b5d1-32cef3990764",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.04311,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 2962.0099579915404,
      "p95_latency_ms": 3591.3396447896957,
      "input_tokens": 6840,
      "output_tokens": 297,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "58494ae4-0b86-47c6-8d3b-0910f5bb8e12",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.07019,
      "cost_per_success_usd": 0.023396666666666666,
      "octane_score": 0.49380253597378554,
      "ci_low": 0.4012400354295836,
      "ci_high": 0.5686361617059086,
      "ci_metric": "octane_score",
      "median_latency_ms": 5211.672916077077,
      "p95_latency_ms": 5543.712165672332,
      "input_tokens": 7588,
      "output_tokens": 153,
      "reasoning_tokens": 922,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b5fb7983-18e5-4f7e-9bb3-67f688f937ac",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.026403,
      "cost_per_success_usd": 0.008801,
      "octane_score": 0.2711434306707571,
      "ci_low": 0.18198353872038564,
      "ci_high": 0.4161053507908878,
      "ci_metric": "octane_score",
      "median_latency_ms": 3315.8230409026146,
      "p95_latency_ms": 4084.989741630852,
      "input_tokens": 6024,
      "output_tokens": 315,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b4846a5c-1724-427a-99a7-3be0483d6632",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.002088,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 839.0337498858571,
      "p95_latency_ms": 839.9955498054624,
      "input_tokens": 540,
      "output_tokens": 84,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "d496ddb4-24b5-4956-954a-e7d2094d2b15",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0042120000000000005,
      "cost_per_success_usd": 0.001404,
      "octane_score": 17.190170940170937,
      "ci_low": 9.641839378238345,
      "ci_high": 28.79959239130435,
      "ci_metric": "octane_score",
      "median_latency_ms": 2617.545874789357,
      "p95_latency_ms": 2751.1790381744504,
      "input_tokens": 390,
      "output_tokens": 66,
      "reasoning_tokens": 220,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3300ce92-6116-495a-a5fe-86ca6fee63a0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003852,
      "cost_per_success_usd": 0.001284,
      "octane_score": 8.997923156801663,
      "ci_low": 7.167156241265023,
      "ci_high": 10.15873015873016,
      "ci_metric": "octane_score",
      "median_latency_ms": 2576.5145840123296,
      "p95_latency_ms": 2722.8908092714846,
      "input_tokens": 552,
      "output_tokens": 36,
      "reasoning_tokens": 193,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ce85cbfb-3c88-4373-bdbb-084217f07f1b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002478,
      "cost_per_success_usd": 0.000826,
      "octane_score": 5.148910411622276,
      "ci_low": 4.7150776053215075,
      "ci_high": 5.57161125319693,
      "ci_metric": "octane_score",
      "median_latency_ms": 1430.0003750249743,
      "p95_latency_ms": 2254.0823373012245,
      "input_tokens": 507,
      "output_tokens": 24,
      "reasoning_tokens": 98,
      "undefined_reason": null
    },
    {
      "snapshot_id": "14e883d3-ce61-492b-88c8-508af8fd52db",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001758,
      "cost_per_success_usd": 0.000586,
      "octane_score": 4.072241183162684,
      "ci_low": 3.4146757679180886,
      "ci_high": 5.271331058020477,
      "ci_metric": "octane_score",
      "median_latency_ms": 778.7990821525455,
      "p95_latency_ms": 2234.388770442456,
      "input_tokens": 339,
      "output_tokens": 90,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f0565559-fc6d-493a-975b-ac374550f2ae",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.013305,
      "cost_per_success_usd": 0.004435000000000001,
      "octane_score": 2.605035700864337,
      "ci_low": 2.09803164801235,
      "ci_high": 2.981829476626806,
      "ci_metric": "octane_score",
      "median_latency_ms": 4016.031375154853,
      "p95_latency_ms": 4705.180200375617,
      "input_tokens": 7377,
      "output_tokens": 159,
      "reasoning_tokens": 829,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8d3a5f30-31c2-46d5-9b9b-cf0d9717226d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.012642,
      "cost_per_success_usd": 0.004214,
      "octane_score": 5.727337446606549,
      "ci_low": 3.4181080667381,
      "ci_high": 9.831445302236325,
      "ci_metric": "octane_score",
      "median_latency_ms": 4206.9923328235745,
      "p95_latency_ms": 4981.487483158708,
      "input_tokens": 6696,
      "output_tokens": 249,
      "reasoning_tokens": 742,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e45352bb-2c8f-4762-baae-60889758f9d5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009552000000000001,
      "cost_per_success_usd": 0.0031840000000000006,
      "octane_score": 1.3357412060301503,
      "ci_low": 1.2704145728643212,
      "ci_high": 1.3942399999999997,
      "ci_metric": "octane_score",
      "median_latency_ms": 2582.603874616325,
      "p95_latency_ms": 3119.3982631899416,
      "input_tokens": 6714,
      "output_tokens": 123,
      "reasoning_tokens": 350,
      "undefined_reason": null
    },
    {
      "snapshot_id": "83120fba-3d5e-453b-b800-cf75a19f1a2d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.006870599999999999,
      "cost_per_success_usd": 0.0022901999999999996,
      "octane_score": 1.0419759555206243,
      "ci_low": 0.693083024488241,
      "ci_high": 1.623457154668399,
      "ci_metric": "octane_score",
      "median_latency_ms": 2311.548958532512,
      "p95_latency_ms": 3198.345658089965,
      "input_tokens": 6024,
      "output_tokens": 321,
      "reasoning_tokens": 259,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cb30a571-0921-42f4-955e-9844516c445e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0064548,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 2014.1095845028758,
      "p95_latency_ms": 2350.822045188397,
      "input_tokens": 6840,
      "output_tokens": 303,
      "reasoning_tokens": 93,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "ff43708f-84b0-4c9f-8c70-8a6667e0265b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00012759,
      "cost_per_success_usd": 4.2529999999999994e-05,
      "octane_score": 100.00000000000001,
      "ci_low": 4.045e-05,
      "ci_high": 4.357e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 4318.186417222023,
      "p95_latency_ms": 4418.063955008984,
      "input_tokens": 666,
      "output_tokens": 11,
      "reasoning_tokens": 622,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c7f447db-fbb3-43c7-8846-fd738f930dec",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 7.159e-05,
      "cost_per_success_usd": 2.3863333333333332e-05,
      "octane_score": 100.0,
      "ci_low": 2.001e-05,
      "ci_high": 3.089e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 5186.99570838362,
      "p95_latency_ms": 6201.015208289027,
      "input_tokens": 522,
      "output_tokens": 115,
      "reasoning_tokens": 214,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8f4aff3f-2847-4892-8d21-b2e719cd7da3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00072405,
      "cost_per_success_usd": 0.00024134999999999998,
      "octane_score": 99.99999999999999,
      "ci_low": 0.00014649000000000004,
      "ci_high": 0.00042393000000000004,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 19312.368124723434,
      "p95_latency_ms": 22562.53628740087,
      "input_tokens": 573,
      "output_tokens": 381,
      "reasoning_tokens": 3777,
      "undefined_reason": null
    },
    {
      "snapshot_id": "af5b3811-ad8f-4443-a4c5-7e835a777cea",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0003466000000000001,
      "cost_per_success_usd": 0.00011553333333333335,
      "octane_score": 100.0,
      "ci_low": 9.059999999999999e-05,
      "ci_high": 0.00012800000000000002,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 13229.666333645582,
      "p95_latency_ms": 13863.062970992178,
      "input_tokens": 696,
      "output_tokens": 294,
      "reasoning_tokens": 1622,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ac9e3f44-f1fb-4f32-883b-e465539a54a9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00012512,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 4532.817333005369,
      "p95_latency_ms": 4673.687096312642,
      "input_tokens": 697,
      "output_tokens": 55,
      "reasoning_tokens": 558,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "3b3d0e5b-a3bf-4fd4-b8e3-f359ea956072",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.009088,
      "cost_per_success_usd": 0.004544,
      "octane_score": null,
      "ci_low": 0.0027083999999999997,
      "ci_high": 0.0098064,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 7492.977083660662,
      "p95_latency_ms": 8501.81052070111,
      "input_tokens": 1140,
      "output_tokens": 94,
      "reasoning_tokens": 1222,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "be571a9f-b49d-4597-bf9c-606bf7eee543",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0032456,
      "cost_per_success_usd": 0.0010818666666666667,
      "octane_score": 2.2057554843480403,
      "ci_low": 1.7772281410826347,
      "ci_high": 2.8174024078803352,
      "ci_metric": "octane_score",
      "median_latency_ms": 2480.4835002869368,
      "p95_latency_ms": 2802.025712840259,
      "input_tokens": 933,
      "output_tokens": 80,
      "reasoning_tokens": 295,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cd6079e9-00b0-4ee9-ab76-ba4b8a08588f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.006628,
      "cost_per_success_usd": 0.0022093333333333335,
      "octane_score": 1.9250150875075436,
      "ci_low": 1.7904200230149598,
      "ci_high": 2.100454365863295,
      "ci_metric": "octane_score",
      "median_latency_ms": 5900.868624448776,
      "p95_latency_ms": 6051.355037465692,
      "input_tokens": 1098,
      "output_tokens": 6,
      "reasoning_tokens": 914,
      "undefined_reason": null
    },
    {
      "snapshot_id": "932d1b0a-ef38-4f6f-95f0-78fd6816c7f6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.015994,
      "cost_per_success_usd": 0.005331333333333334,
      "octane_score": 4.527010128798298,
      "ci_low": 2.4201257180015174,
      "ci_high": 7.871156607416944,
      "ci_metric": "octane_score",
      "median_latency_ms": 12712.894333526492,
      "p95_latency_ms": 17132.155370432884,
      "input_tokens": 1002,
      "output_tokens": 60,
      "reasoning_tokens": 2453,
      "undefined_reason": null
    },
    {
      "snapshot_id": "325885d6-ae4b-4c06-87ae-81716df2a163",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0119216,
      "cost_per_success_usd": 0.003973866666666667,
      "octane_score": 2.90732787545296,
      "ci_low": 2.4132871280946584,
      "ci_high": 3.479773814702045,
      "ci_metric": "octane_score",
      "median_latency_ms": 9574.90804232657,
      "p95_latency_ms": 10384.919478930531,
      "input_tokens": 1146,
      "output_tokens": 18,
      "reasoning_tokens": 1732,
      "undefined_reason": null
    },
    {
      "snapshot_id": "97a78104-0e4c-4fa4-a58b-7b6c157a967d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.027000000000000003,
      "cost_per_success_usd": 0.009000000000000001,
      "octane_score": 1.2837037037037038,
      "ci_low": 1.0425417327974282,
      "ci_high": 1.4749942383037569,
      "ci_metric": "octane_score",
      "median_latency_ms": 23800.923041068017,
      "p95_latency_ms": 27688.04374197498,
      "input_tokens": 1146,
      "output_tokens": 18,
      "reasoning_tokens": 4260,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2c78b45c-fd97-4fea-90da-32844a86e6cf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.023856000000000002,
      "cost_per_success_usd": 0.007952,
      "octane_score": 3.03508551307847,
      "ci_low": 1.6631845172641304,
      "ci_high": 5.331111670020121,
      "ci_metric": "octane_score",
      "median_latency_ms": 18503.356290981174,
      "p95_latency_ms": 24416.446178779002,
      "input_tokens": 1002,
      "output_tokens": 60,
      "reasoning_tokens": 3710,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d8f045d9-9fa9-4557-af50-c4dc76fc1b26",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009966000000000001,
      "cost_per_success_usd": 0.0033220000000000003,
      "octane_score": 1.2802528597230582,
      "ci_low": 1.0068655303030303,
      "ci_high": 1.6332565284178187,
      "ci_metric": "octane_score",
      "median_latency_ms": 7627.075208351016,
      "p95_latency_ms": 10741.319558396934,
      "input_tokens": 1098,
      "output_tokens": 6,
      "reasoning_tokens": 1449,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9132acb5-0b39-4755-8856-dd6231787bc6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00732,
      "cost_per_success_usd": 0.00244,
      "octane_score": 0.9780054644808743,
      "ci_low": 0.7444565744784091,
      "ci_high": 1.2893996247654784,
      "ci_metric": "octane_score",
      "median_latency_ms": 5283.120124600828,
      "p95_latency_ms": 6077.910612616688,
      "input_tokens": 933,
      "output_tokens": 66,
      "reasoning_tokens": 1003,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3d04f48b-c61c-46aa-8e41-d08994b5f4fe",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.018408,
      "cost_per_success_usd": 0.018408,
      "octane_score": null,
      "ci_low": 0.006897999999999999,
      "ci_high": 0.019698,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 12190.569249913096,
      "p95_latency_ms": 12957.609437499195,
      "input_tokens": 1140,
      "output_tokens": 75,
      "reasoning_tokens": 2773,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "927e0b56-9020-4e51-9607-4bb273ebf262",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00395858,
      "cost_per_success_usd": 0.0013195266666666667,
      "octane_score": 1.8084767770261052,
      "ci_low": 1.276493106022328,
      "ci_high": 3.023210204986527,
      "ci_metric": "octane_score",
      "median_latency_ms": 11915.389833040535,
      "p95_latency_ms": 12217.625657655299,
      "input_tokens": 336,
      "output_tokens": 78,
      "reasoning_tokens": 1652,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5f70b97e-bd4f-470d-8b4f-0b21d969dc1b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.008905139999999999,
      "cost_per_success_usd": 0.0029683799999999996,
      "octane_score": 1.432768041827529,
      "ci_low": 1.1566651708617195,
      "ci_high": 2.0101903843608797,
      "ci_metric": "octane_score",
      "median_latency_ms": 20116.75270833075,
      "p95_latency_ms": 22183.336196653545,
      "input_tokens": 513,
      "output_tokens": 12,
      "reasoning_tokens": 3901,
      "undefined_reason": null
    },
    {
      "snapshot_id": "af224ec6-1549-4511-9595-990e25f7ee69",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.01493542,
      "cost_per_success_usd": 0.004978473333333334,
      "octane_score": 4.84787170364141,
      "ci_low": 2.620080528138943,
      "ci_high": 8.51526103718543,
      "ci_metric": "octane_score",
      "median_latency_ms": 49873.05787485093,
      "p95_latency_ms": 54186.70646240935,
      "input_tokens": 396,
      "output_tokens": 1168,
      "reasoning_tokens": 5434,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ac52f92a-3252-4268-860c-c11cff7e380a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.01907765,
      "cost_per_success_usd": 0.006359216666666667,
      "octane_score": 1.8167856103870237,
      "ci_low": 1.2725687273394835,
      "ci_high": 2.620114152369191,
      "ci_metric": "octane_score",
      "median_latency_ms": 42971.18175029755,
      "p95_latency_ms": 58508.439975231886,
      "input_tokens": 556,
      "output_tokens": 150,
      "reasoning_tokens": 8249,
      "undefined_reason": null
    },
    {
      "snapshot_id": "25265752-5914-44fd-a5da-659b46007db3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.009368379999999999,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 19815.147291868925,
      "p95_latency_ms": 20517.96959117055,
      "input_tokens": 546,
      "output_tokens": 86,
      "reasoning_tokens": 4030,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "b20b7798-05a3-4937-969a-cd2f2b244e5e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0147264,
      "cost_per_success_usd": 0.0049088000000000005,
      "octane_score": 0.8664031942633637,
      "ci_low": 0.8240303129074315,
      "ci_high": 0.9051437593485124,
      "ci_metric": "octane_score",
      "median_latency_ms": 2619.624917395413,
      "p95_latency_ms": 2752.2290797904134,
      "input_tokens": 747,
      "output_tokens": 0,
      "reasoning_tokens": 913,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0569ffc6-e80b-4049-b743-e6f80585a73a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0112302,
      "cost_per_success_usd": 0.0037433999999999996,
      "octane_score": 0.6374775159836868,
      "ci_low": 0.49204110743694485,
      "ci_high": 0.8314526786621939,
      "ci_metric": "octane_score",
      "median_latency_ms": 1620.5494161695242,
      "p95_latency_ms": 2075.207703281194,
      "input_tokens": 446,
      "output_tokens": 408,
      "reasoning_tokens": 263,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1b1aabda-7f37-429e-8127-a0e9c8e59538",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0074298,
      "cost_per_success_usd": 0.0024766000000000002,
      "octane_score": null,
      "ci_low": 0.0022788,
      "ci_high": 0.0027972000000000006,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 2097.3422080278397,
      "p95_latency_ms": 2134.3742466531694,
      "input_tokens": 780,
      "output_tokens": 113,
      "reasoning_tokens": 330,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "2306ef85-c8ed-4b19-ac9c-13df764157d6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": 3,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.035116800000000004,
      "cost_per_success_usd": 0.035116800000000004,
      "octane_score": 0.3289973270153697,
      "ci_low": 0.2663551723900294,
      "ci_high": 1.094812119374321,
      "ci_metric": "octane_score",
      "median_latency_ms": 5646.092458628118,
      "p95_latency_ms": 5803.1805214472115,
      "input_tokens": 789,
      "output_tokens": 490,
      "reasoning_tokens": 1797,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b5f40a79-86b3-4a8f-b34e-f6893347dba2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0365514,
      "cost_per_success_usd": 0.0121838,
      "octane_score": 1.9809090759861452,
      "ci_low": 0.9570361933687676,
      "ci_high": 3.8982437292727363,
      "ci_metric": "octane_score",
      "median_latency_ms": 3337.8211250528693,
      "p95_latency_ms": 3927.9881378635764,
      "input_tokens": 627,
      "output_tokens": 521,
      "reasoning_tokens": 1871,
      "undefined_reason": null
    },
    {
      "snapshot_id": "66133bdb-f27f-4056-ae01-d237b6b4f2e3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.00018480000000000002,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 580.4285844787955,
      "p95_latency_ms": 614.0862960368395,
      "input_tokens": 552,
      "output_tokens": 93,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "edc9578e-ed04-40f7-88ab-ec942d004808",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0006902,
      "cost_per_success_usd": 0.0003451,
      "octane_score": 12.323964068385974,
      "ci_low": 6.0675636150921335,
      "ci_high": 19.01024030005309,
      "ci_metric": "octane_score",
      "median_latency_ms": 2874.937292188406,
      "p95_latency_ms": 6763.3070416748515,
      "input_tokens": 531,
      "output_tokens": 730,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ccb57b96-1c64-4727-8246-80dcd5c8afe0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.000135,
      "cost_per_success_usd": 4.5e-05,
      "octane_score": 53.029629629629625,
      "ci_low": 44.46666666666667,
      "ci_high": 68.64444444444443,
      "ci_metric": "octane_score",
      "median_latency_ms": 1281.4677087590098,
      "p95_latency_ms": 1331.9973079487681,
      "input_tokens": 351,
      "output_tokens": 81,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "78bd204c-97dd-4e14-80bb-32b7ad02b364",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0008174,
      "cost_per_success_usd": 0.00040869999999999996,
      "octane_score": 59.05309517983851,
      "ci_low": 20.166621511785426,
      "ci_high": 112.50168501460347,
      "ci_metric": "octane_score",
      "median_latency_ms": 3835.215917788446,
      "p95_latency_ms": 16503.113716375083,
      "input_tokens": 399,
      "output_tokens": 922,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c8205e45-37cf-4fe5-9b5a-e5559b46ffcf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 1,
      "success_rate": 0.0,
      "total_cost_usd": 0.0009436000000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3022.982334718108,
      "p95_latency_ms": 13922.01488362625,
      "input_tokens": 570,
      "output_tokens": 1037,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "cfd0b7f0-c24d-4d64-aa2f-33df352ce2a1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.675209127366543,
      "p95_latency_ms": 33.52006981149316,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "b0486406-7ccf-44e0-925f-61e5c9b2c08d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 31.701000407338142,
      "p95_latency_ms": 34.806074760854244,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "6194107d-b81d-415f-9bc0-a3195bb2546e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.41104134172201,
      "p95_latency_ms": 32.931679114699364,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "3f3d6959-738d-4503-bb8a-a23dbb609a28",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 29.42450065165758,
      "p95_latency_ms": 40.00489991158246,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "e5b8a070-2b84-4c6d-8b59-e5250a0903c9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 27.570624835789204,
      "p95_latency_ms": 30.290238093584776,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "a7b7db0d-c962-4239-8feb-a87674f388b5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.02098,
      "cost_per_success_usd": 0.006993333333333333,
      "octane_score": null,
      "ci_low": 0.006660000000000001,
      "ci_high": 0.00766,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 1447.1089579164982,
      "p95_latency_ms": 2096.329883299768,
      "input_tokens": 888,
      "output_tokens": 242,
      "reasoning_tokens": 0,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "261b0a19-2b05-43eb-bf39-98bc8348ac93",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.01245,
      "cost_per_success_usd": 0.00415,
      "octane_score": 0.5750200803212852,
      "ci_low": 0.4821686746987952,
      "ci_high": 0.7443373493975903,
      "ci_metric": "octane_score",
      "median_latency_ms": 1104.936582967639,
      "p95_latency_ms": 1173.0285961180925,
      "input_tokens": 540,
      "output_tokens": 141,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6cfbea92-6aef-4e60-9a84-72d7fba177e8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.07157,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 4014.8799577727914,
      "p95_latency_ms": 4065.585283562541,
      "input_tokens": 807,
      "output_tokens": 329,
      "reasoning_tokens": 941,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "4d152fec-067a-43a7-b715-3c9dba58831d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.08665,
      "cost_per_success_usd": 0.028883333333333334,
      "octane_score": 0.835603000577034,
      "ci_low": 0.46484114977307117,
      "ci_high": 1.4379609544468546,
      "ci_metric": "octane_score",
      "median_latency_ms": 5114.379666745663,
      "p95_latency_ms": 5436.538941599429,
      "input_tokens": 585,
      "output_tokens": 411,
      "reasoning_tokens": 1205,
      "undefined_reason": null
    },
    {
      "snapshot_id": "402c115f-517b-4419-a593-1720720f3b07",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 3,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.026029999999999998,
      "cost_per_success_usd": 0.013014999999999999,
      "octane_score": 0.32677679600461007,
      "ci_low": 0.12355568790946996,
      "ci_high": 0.7073051948051948,
      "ci_metric": "octane_score",
      "median_latency_ms": 3163.3627079427242,
      "p95_latency_ms": 3402.454907540232,
      "input_tokens": 828,
      "output_tokens": 142,
      "reasoning_tokens": 213,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2a563932-5fd0-46a2-ab0b-ef9e410a382f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00567905,
      "cost_per_success_usd": 0.0018930166666666665,
      "octane_score": 6.1031334466151925,
      "ci_low": 4.8091299155128935,
      "ci_high": 7.19625345054731,
      "ci_metric": "octane_score",
      "median_latency_ms": 32614.780915901065,
      "p95_latency_ms": 33398.5080787912,
      "input_tokens": 618,
      "output_tokens": 21688,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "79d18024-cbdd-4aa4-a407-e0657ba202a0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00113542,
      "cost_per_success_usd": 0.00056771,
      "octane_score": 4.203437200918309,
      "ci_low": 2.098499146220904,
      "ci_high": 6.312284730195179,
      "ci_metric": "octane_score",
      "median_latency_ms": 10213.242458179593,
      "p95_latency_ms": 10379.76645855233,
      "input_tokens": 360,
      "output_tokens": 4277,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "058f8780-060a-4372-83ed-1f910e49eb01",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.002217865,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 9179.986667819321,
      "p95_latency_ms": 20434.795217309143,
      "input_tokens": 429,
      "output_tokens": 8423,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "cdfdd85d-074d-4788-ab26-0855e174554e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0008873150000000001,
      "cost_per_success_usd": 0.0002957716666666667,
      "octane_score": 14.379335410761678,
      "ci_low": 12.2132985325791,
      "ci_high": 19.99106912030835,
      "ci_metric": "octane_score",
      "median_latency_ms": 6959.113791584969,
      "p95_latency_ms": 7206.305879354477,
      "input_tokens": 579,
      "output_tokens": 3268,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e7b302c7-e515-40cd-8b5a-6c8bb99b0404",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0027424800000000003,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 25247.801499441266,
      "p95_latency_ms": 26034.009624365717,
      "input_tokens": 588,
      "output_tokens": 10401,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "7d2b0093-ea55-40e6-ae68-632657f34939",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007013999999999999,
      "cost_per_success_usd": 0.002338,
      "octane_score": 1.819076133447391,
      "ci_low": 1.6587363494539782,
      "ci_high": 1.9768602540834848,
      "ci_metric": "octane_score",
      "median_latency_ms": 3072.106500156224,
      "p95_latency_ms": 6123.015937581657,
      "input_tokens": 714,
      "output_tokens": 20,
      "reasoning_tokens": 911,
      "undefined_reason": null
    },
    {
      "snapshot_id": "23c09f22-bbe7-41c9-92ca-dfff45690cfa",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.044784,
      "cost_per_success_usd": 0.014927999999999999,
      "octane_score": 0.7739371204001432,
      "ci_low": 0.3514001257172984,
      "ci_high": 2.233856893542758,
      "ci_metric": "octane_score",
      "median_latency_ms": 15401.284916326404,
      "p95_latency_ms": 104043.35966678335,
      "input_tokens": 762,
      "output_tokens": 41,
      "reasoning_tokens": 7169,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b42a9367-253e-414a-b5ef-213bf04ddbcb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": null,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0036839999999999998,
      "cost_per_success_usd": 0.0018419999999999999,
      "octane_score": 1.2955121245023524,
      "ci_low": 0.5691126279863482,
      "ci_high": 2.1714060031595577,
      "ci_metric": "octane_score",
      "median_latency_ms": 3760.4480423033237,
      "p95_latency_ms": 5236.943041998892,
      "input_tokens": 504,
      "output_tokens": 78,
      "reasoning_tokens": 368,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0ae2f349-35ac-4958-82a2-ed6cebf05b29",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.02319,
      "cost_per_success_usd": 0.007729999999999999,
      "octane_score": null,
      "ci_low": 0.002642,
      "ci_high": 0.010274,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 6740.600375458598,
      "p95_latency_ms": 6772.227612044662,
      "input_tokens": 714,
      "output_tokens": 24,
      "reasoning_tokens": 3603,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "12d499d3-dfc0-48a4-b8e1-0a1a80b4fc18",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.011412,
      "cost_per_success_usd": 0.003804,
      "octane_score": 6.344637223974763,
      "ci_low": 3.7555499495459133,
      "ci_high": 11.13261554621849,
      "ci_metric": "octane_score",
      "median_latency_ms": 11884.739791974425,
      "p95_latency_ms": 13261.090329661965,
      "input_tokens": 582,
      "output_tokens": 79,
      "reasoning_tokens": 1629,
      "undefined_reason": null
    },
    {
      "snapshot_id": "65da0bad-a3b6-4627-b4f4-c3b5b6696d9f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0025515,
      "cost_per_success_usd": 0.0008505,
      "octane_score": 2.805800509504213,
      "ci_low": 2.1550530687586527,
      "ci_high": 3.6319811875367427,
      "ci_metric": "octane_score",
      "median_latency_ms": 4183.1602500751615,
      "p95_latency_ms": 4221.723750419915,
      "input_tokens": 486,
      "output_tokens": 449,
      "reasoning_tokens": 280,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7376184a-96f0-4554-9665-70ad9f3af67d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0035396499999999997,
      "cost_per_success_usd": 0.0011798833333333332,
      "octane_score": 3.6045936745158422,
      "ci_low": 3.4296009928582603,
      "ci_high": 3.740074681316795,
      "ci_metric": "octane_score",
      "median_latency_ms": 8335.8334582299,
      "p95_latency_ms": 18111.681046243757,
      "input_tokens": 705,
      "output_tokens": 0,
      "reasoning_tokens": 1007,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bd7770c6-1e26-4207-9355-d1b472213804",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0049773000000000005,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 6873.352541588247,
      "p95_latency_ms": 7746.8104916624725,
      "input_tokens": 714,
      "output_tokens": 6,
      "reasoning_tokens": 1449,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "37a8462e-c96e-4814-8791-9e9334df1009",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00527055,
      "cost_per_success_usd": 0.00175685,
      "octane_score": 13.737655462902353,
      "ci_low": 6.905610489613212,
      "ci_high": 24.583982816814977,
      "ci_metric": "octane_score",
      "median_latency_ms": 9039.395458996296,
      "p95_latency_ms": 19801.210370566692,
      "input_tokens": 555,
      "output_tokens": 457,
      "reasoning_tokens": 1112,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7e96e671-0a36-42af-a7fa-4dd5b5472e6e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0133442,
      "cost_per_success_usd": 0.004448066666666667,
      "octane_score": 2.5973831327468115,
      "ci_low": 2.0848504463683684,
      "ci_high": 3.0530466861722427,
      "ci_metric": "octane_score",
      "median_latency_ms": 18253.14279086888,
      "p95_latency_ms": 20039.885528571904,
      "input_tokens": 740,
      "output_tokens": 614,
      "reasoning_tokens": 3452,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e9f6c066-06bf-481e-ad25-209f32f0eb0a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.10570199999999999,
      "cost_per_success_usd": 0.035233999999999994,
      "octane_score": 0.32790297250761585,
      "ci_low": 0.22020610172937402,
      "ci_high": 0.4901209986215348,
      "ci_metric": "octane_score",
      "median_latency_ms": 80953.91295943409,
      "p95_latency_ms": 141060.54788390175,
      "input_tokens": 732,
      "output_tokens": 45,
      "reasoning_tokens": 17328,
      "undefined_reason": null
    },
    {
      "snapshot_id": "75280d4e-2629-454d-94ed-9aaf586fb16b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.010272,
      "cost_per_success_usd": 0.003424,
      "octane_score": 7.048773364485981,
      "ci_low": 3.9741057127602772,
      "ci_high": 12.259398496240602,
      "ci_metric": "octane_score",
      "median_latency_ms": 9149.834458716214,
      "p95_latency_ms": 11413.835120387374,
      "input_tokens": 543,
      "output_tokens": 81,
      "reasoning_tokens": 1450,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f8a3debf-c1b6-4dfc-9ebd-96f58a147939",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.006498,
      "cost_per_success_usd": 0.002166,
      "octane_score": 1.9635272391505079,
      "ci_low": 1.791450777202073,
      "ci_high": 2.2646432374866876,
      "ci_metric": "octane_score",
      "median_latency_ms": 6559.1824585571885,
      "p95_latency_ms": 6869.541008677334,
      "input_tokens": 693,
      "output_tokens": 33,
      "reasoning_tokens": 819,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ec0c4927-2939-41a9-bd50-315e015f6fc1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004554000000000001,
      "cost_per_success_usd": 0.0015180000000000003,
      "octane_score": 1.5720245937637238,
      "ci_low": 1.19602048857368,
      "ci_high": 2.045386904761904,
      "ci_metric": "octane_score",
      "median_latency_ms": 4498.57116676867,
      "p95_latency_ms": 4890.7921416684985,
      "input_tokens": 474,
      "output_tokens": 124,
      "reasoning_tokens": 477,
      "undefined_reason": null
    },
    {
      "snapshot_id": "28cb89e3-95e6-4bce-b7cd-3334d5dd10f5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.037722,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 39887.57916726172,
      "p95_latency_ms": 47835.903291776776,
      "input_tokens": 702,
      "output_tokens": 103,
      "reasoning_tokens": 5950,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "d8784e44-0de7-43a5-8f93-ac2ca2a602db",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.010368,
      "cost_per_success_usd": 0.010368,
      "octane_score": 0.23016332304526746,
      "ci_low": 0.17620365565357787,
      "ci_high": 0.9772044772044771,
      "ci_metric": "octane_score",
      "median_latency_ms": 16416.110957972705,
      "p95_latency_ms": 21087.952683307227,
      "input_tokens": 432,
      "output_tokens": 85,
      "reasoning_tokens": 3299,
      "undefined_reason": null
    },
    {
      "snapshot_id": "366c1739-b039-4520-9201-23e4f434c1d0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": null,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.0292605,
      "cost_per_success_usd": 0.0292605,
      "octane_score": null,
      "ci_low": 0.0173665,
      "ci_high": 0.0313395,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 31056.352541781962,
      "p95_latency_ms": 74532.97174144536,
      "input_tokens": 663,
      "output_tokens": 87,
      "reasoning_tokens": 9556,
      "undefined_reason": "baseline_undefined"
    },
    {
      "snapshot_id": "0716fe3a-c8a5-458b-b842-d2eeb41c0f59",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.036865499999999995,
      "cost_per_success_usd": 0.012288499999999999,
      "octane_score": 1.96403141148228,
      "ci_low": 1.0134342859057257,
      "ci_high": 3.708761646472158,
      "ci_metric": "octane_score",
      "median_latency_ms": 53632.12858326733,
      "p95_latency_ms": 75758.00165878609,
      "input_tokens": 507,
      "output_tokens": 72,
      "reasoning_tokens": 12132,
      "undefined_reason": null
    },
    {
      "snapshot_id": "07d617d7-371e-4245-93d7-322b15ee2ad2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0396135,
      "cost_per_success_usd": 0.013204500000000001,
      "octane_score": 0.32208716725358777,
      "ci_low": 0.25419119624660064,
      "ci_high": 0.38542752276949577,
      "ci_metric": "octane_score",
      "median_latency_ms": 54913.87966554612,
      "p95_latency_ms": 75063.98540418595,
      "input_tokens": 651,
      "output_tokens": 18,
      "reasoning_tokens": 13078,
      "undefined_reason": null
    },
    {
      "snapshot_id": "63c96bd5-a459-4873-9e6d-580309fea061",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.087798,
      "cost_per_success_usd": 0.043899,
      "octane_score": 0.26317987501613554,
      "ci_low": 0.08400302106595814,
      "ci_high": 0.8039338482592259,
      "ci_metric": "octane_score",
      "median_latency_ms": 113601.28758382052,
      "p95_latency_ms": 214507.19864517447,
      "input_tokens": 690,
      "output_tokens": 274,
      "reasoning_tokens": 28877,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4767daf1-6fba-4f87-beed-f3b0552ba56f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-notice-period-calculation",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.028025999999999995,
      "cost_per_success_usd": 0.009341999999999998,
      "octane_score": 2.5834938985228004,
      "ci_low": 1.4194308969348541,
      "ci_high": 4.450552911534155,
      "ci_metric": "octane_score",
      "median_latency_ms": 47660.362291149795,
      "p95_latency_ms": 64165.94859128818,
      "input_tokens": 552,
      "output_tokens": 72,
      "reasoning_tokens": 11028,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6b5f0070-f8cb-4cab-bad9-2965d7a0040d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-liability-cap-carveout",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.031897999999999996,
      "cost_per_success_usd": 0.010632666666666665,
      "octane_score": 0.399993730014421,
      "ci_low": 0.3346579061258213,
      "ci_high": 0.5168621255392842,
      "ci_metric": "octane_score",
      "median_latency_ms": 59386.676042340696,
      "p95_latency_ms": 67877.61612916365,
      "input_tokens": 696,
      "output_tokens": 18,
      "reasoning_tokens": 12602,
      "undefined_reason": null
    },
    {
      "snapshot_id": "480a4dd5-d1d1-4384-a879-cbb35341b2b1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-governing-law-extraction",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.020026,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 38497.75270745158,
      "p95_latency_ms": 47710.08513299748,
      "input_tokens": 477,
      "output_tokens": 91,
      "reasoning_tokens": 7824,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "24ef6437-9bd1-44d0-b525-d19919f290ce",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-clause-precedence",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0141015,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 18019.357125274837,
      "p95_latency_ms": 54404.05192561447,
      "input_tokens": 708,
      "output_tokens": 87,
      "reasoning_tokens": 5412,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "81aadc26-1cde-407f-ab57-3326d88f224b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "legal",
      "task_id": "legal-termination-earliest-date",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.110405,
      "cost_per_success_usd": 0.0552025,
      "octane_score": 0.20929003819271474,
      "ci_low": 0.08505604841482704,
      "ci_high": 0.38540673458122154,
      "ci_metric": "octane_score",
      "median_latency_ms": 202100.87654273957,
      "p95_latency_ms": 223371.73780379817,
      "input_tokens": 735,
      "output_tokens": 20,
      "reasoning_tokens": 43995,
      "undefined_reason": null
    },
    {
      "snapshot_id": "01448479-454b-43eb-a389-c0ca202970c9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000603,
      "cost_per_success_usd": 0.000201,
      "octane_score": 10.427860696517413,
      "ci_low": 10.427860696517413,
      "ci_high": 10.427860696517413,
      "ci_metric": "octane_score",
      "median_latency_ms": 508.3548752591014,
      "p95_latency_ms": 534.4158005900681,
      "input_tokens": 528,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b67ca947-c4fe-4283-bafa-0c77298b6fba",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.000411,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 654.276791960001,
      "p95_latency_ms": 688.1333289667964,
      "input_tokens": 306,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "310a8c7d-58ab-4fe3-9a16-07756c3e8b36",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00039600000000000003,
      "cost_per_success_usd": 0.000132,
      "octane_score": 13.376262626262625,
      "ci_low": 13.204545454545455,
      "ci_high": 13.462121212121213,
      "ci_metric": "octane_score",
      "median_latency_ms": 503.66554223001003,
      "p95_latency_ms": 515.7651036046445,
      "input_tokens": 336,
      "output_tokens": 12,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d72714ef-97c5-485f-b2a5-4836d9fb3db8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0006900000000000001,
      "cost_per_success_usd": 0.00023000000000000003,
      "octane_score": 29.52898550724637,
      "ci_low": 28.856521739130432,
      "ci_high": 30.778260869565212,
      "ci_metric": "octane_score",
      "median_latency_ms": 516.6651662439108,
      "p95_latency_ms": 582.6811043545604,
      "input_tokens": 615,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "967718e1-7d01-4f7b-98c3-927d83fad7de",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000267,
      "cost_per_success_usd": 8.9e-05,
      "octane_score": 19.812734082397004,
      "ci_low": 13.573033707865168,
      "ci_high": 23.887640449438212,
      "ci_metric": "octane_score",
      "median_latency_ms": 532.5812501832843,
      "p95_latency_ms": 659.6519877202808,
      "input_tokens": 162,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1b32bc17-28b0-49c2-9d45-c5e843fdb602",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003147,
      "cost_per_success_usd": 0.001049,
      "octane_score": 3.7403876707975856,
      "ci_low": 3.7025738798856054,
      "ci_high": 3.799809342230696,
      "ci_metric": "octane_score",
      "median_latency_ms": 2204.00133356452,
      "p95_latency_ms": 2225.2238212153316,
      "input_tokens": 267,
      "output_tokens": 576,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4ad5a43b-beac-4a8f-bec7-40178c8136f3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001107,
      "cost_per_success_usd": 0.00036899999999999997,
      "octane_score": 6.384823848238482,
      "ci_low": 6.13008130081301,
      "ci_high": 6.880758807588077,
      "ci_metric": "octane_score",
      "median_latency_ms": 726.9021244719625,
      "p95_latency_ms": 751.6569623723626,
      "input_tokens": 327,
      "output_tokens": 156,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a1483ccf-74d5-4d2e-a68e-b0cabe77e6dc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001989,
      "cost_per_success_usd": 0.000663,
      "octane_score": 2.775766716943188,
      "ci_low": 2.0211161387631975,
      "ci_high": 3.5339366515837107,
      "ci_metric": "octane_score",
      "median_latency_ms": 1599.7918341308832,
      "p95_latency_ms": 1624.5240199379623,
      "input_tokens": 354,
      "output_tokens": 327,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9794cada-4c0c-489a-b95b-c541f059237e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0006360000000000001,
      "cost_per_success_usd": 0.00021200000000000003,
      "octane_score": 13.083333333333334,
      "ci_low": 12.523584905660378,
      "ci_high": 13.641509433962263,
      "ci_metric": "octane_score",
      "median_latency_ms": 734.5862500369549,
      "p95_latency_ms": 860.5697876773775,
      "input_tokens": 216,
      "output_tokens": 84,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e8c56779-5c96-4d27-9951-5d69ebee69da",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.01976,
      "cost_per_success_usd": 0.006586666666666667,
      "octane_score": 0.5956983805668017,
      "ci_low": 0.5853523952846751,
      "ci_high": 0.6064399793920661,
      "ci_metric": "octane_score",
      "median_latency_ms": 4678.53937484324,
      "p95_latency_ms": 4710.9401621855795,
      "input_tokens": 267,
      "output_tokens": 737,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "048ea426-b1ab-443e-84c9-a05fce663e6b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0013349999999999998,
      "cost_per_success_usd": 0.0004449999999999999,
      "octane_score": 3.9625468164794015,
      "ci_low": 2.714606741573034,
      "ci_high": 4.777528089887643,
      "ci_metric": "octane_score",
      "median_latency_ms": 2347.1057089045644,
      "p95_latency_ms": 2658.7811456993218,
      "input_tokens": 162,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "68bbaf62-aaaf-44a6-9179-db1798b86973",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0030150000000000003,
      "cost_per_success_usd": 0.001005,
      "octane_score": 2.0855721393034825,
      "ci_low": 2.0855721393034825,
      "ci_high": 2.0855721393034825,
      "ci_metric": "octane_score",
      "median_latency_ms": 1614.2771244049072,
      "p95_latency_ms": 1632.3274875059724,
      "input_tokens": 528,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7d42d19f-6b41-451a-92e6-76e524a778af",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00198,
      "cost_per_success_usd": 0.00066,
      "octane_score": 2.6752525252525254,
      "ci_low": 2.640909090909091,
      "ci_high": 2.692424242424243,
      "ci_metric": "octane_score",
      "median_latency_ms": 1634.4585828483105,
      "p95_latency_ms": 1641.1853333935142,
      "input_tokens": 336,
      "output_tokens": 12,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8d4d355e-b68f-4540-8d8d-9092380722cc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.012205,
      "cost_per_success_usd": 0.0040683333333333335,
      "octane_score": 0.6817697664891439,
      "ci_low": 0.6288603656164918,
      "ci_high": 0.7630444750114626,
      "ci_metric": "octane_score",
      "median_latency_ms": 3717.7297500893474,
      "p95_latency_ms": 3752.634112164378,
      "input_tokens": 216,
      "output_tokens": 445,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ac65091f-2de3-4a3c-83ec-b0eea19ac6e2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002145,
      "cost_per_success_usd": 0.000715,
      "octane_score": 2.5738927738927737,
      "ci_low": 1.8741258741258737,
      "ci_high": 3.276923076923077,
      "ci_metric": "octane_score",
      "median_latency_ms": 1501.1430829763412,
      "p95_latency_ms": 1547.5160703994334,
      "input_tokens": 354,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "742912a7-b7cb-44ef-bf70-2346fd7bdfc4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004710000000000001,
      "cost_per_success_usd": 0.0015700000000000002,
      "octane_score": 1.5006369426751587,
      "ci_low": 1.4407643312101912,
      "ci_high": 1.6171974522292991,
      "ci_metric": "octane_score",
      "median_latency_ms": 2386.4802913740277,
      "p95_latency_ms": 2451.070328615606,
      "input_tokens": 327,
      "output_tokens": 123,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "060b6722-9086-43ca-9a07-beae1414035d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00345,
      "cost_per_success_usd": 0.00115,
      "octane_score": 5.905797101449275,
      "ci_low": 5.771304347826088,
      "ci_high": 6.155652173913044,
      "ci_metric": "octane_score",
      "median_latency_ms": 1785.3194177150726,
      "p95_latency_ms": 1940.0033554062247,
      "input_tokens": 615,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bccdec5d-5ce1-4deb-92da-76805c502647",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001905,
      "cost_per_success_usd": 0.000635,
      "octane_score": 3.3942257217847773,
      "ci_low": 3.283464566929134,
      "ci_high": 3.489763779527559,
      "ci_metric": "octane_score",
      "median_latency_ms": 1671.3433749973774,
      "p95_latency_ms": 1836.0319003462791,
      "input_tokens": 306,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "939fcf77-98a7-43a6-85b2-f6539f5e9c7e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000801,
      "cost_per_success_usd": 0.000267,
      "octane_score": 6.604244694132335,
      "ci_low": 4.524344569288389,
      "ci_high": 7.962546816479404,
      "ci_metric": "octane_score",
      "median_latency_ms": 1676.8547501415014,
      "p95_latency_ms": 1913.9635751955211,
      "input_tokens": 162,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9362f4ad-a920-40e1-9f13-7d5fa1474e4b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001809,
      "cost_per_success_usd": 0.000603,
      "octane_score": 3.4759535655058045,
      "ci_low": 3.4759535655058045,
      "ci_high": 3.4759535655058045,
      "ci_metric": "octane_score",
      "median_latency_ms": 1758.6637074127793,
      "p95_latency_ms": 2038.144046254456,
      "input_tokens": 528,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d663d786-4924-4dc3-afef-a6834a9025f3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0019080000000000002,
      "cost_per_success_usd": 0.0006360000000000001,
      "octane_score": 4.361111111111112,
      "ci_low": 4.174528301886793,
      "ci_high": 4.547169811320754,
      "ci_metric": "octane_score",
      "median_latency_ms": 1484.0929582715034,
      "p95_latency_ms": 2435.0982455536723,
      "input_tokens": 216,
      "output_tokens": 84,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8b73265b-4d39-481d-95d7-2cdb3cf84e78",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001287,
      "cost_per_success_usd": 0.00042899999999999997,
      "octane_score": 4.28982128982129,
      "ci_low": 3.1235431235431235,
      "ci_high": 5.461538461538462,
      "ci_metric": "octane_score",
      "median_latency_ms": 1938.5071657598019,
      "p95_latency_ms": 4325.594866834581,
      "input_tokens": 354,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5b9af757-1783-43e3-a679-929b2d94153c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001143,
      "cost_per_success_usd": 0.000381,
      "octane_score": 5.657042869641296,
      "ci_low": 5.47244094488189,
      "ci_high": 5.816272965879265,
      "ci_metric": "octane_score",
      "median_latency_ms": 1026.9060423597693,
      "p95_latency_ms": 1061.577566061169,
      "input_tokens": 306,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f64a0bbd-bee5-43da-8282-327c098aa422",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0033209999999999997,
      "cost_per_success_usd": 0.001107,
      "octane_score": 2.128274616079494,
      "ci_low": 2.0433604336043367,
      "ci_high": 2.2935862691960254,
      "ci_metric": "octane_score",
      "median_latency_ms": 1935.0338745862246,
      "p95_latency_ms": 2449.0387627854943,
      "input_tokens": 327,
      "output_tokens": 156,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "104b7a9f-73e8-40de-bf01-6e8340a780da",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0011879999999999998,
      "cost_per_success_usd": 0.0003959999999999999,
      "octane_score": 4.458754208754209,
      "ci_low": 4.401515151515153,
      "ci_high": 4.487373737373739,
      "ci_metric": "octane_score",
      "median_latency_ms": 1007.549375295639,
      "p95_latency_ms": 1012.6845507882535,
      "input_tokens": 336,
      "output_tokens": 12,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fc648748-bffc-4f5b-aaaf-08bacdd33ebe",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0068850000000000005,
      "cost_per_success_usd": 0.002295,
      "octane_score": 2.959331880900508,
      "ci_low": 2.891938997821351,
      "ci_high": 3.084531590413943,
      "ci_metric": "octane_score",
      "median_latency_ms": 2872.519291937351,
      "p95_latency_ms": 2979.5139925554395,
      "input_tokens": 615,
      "output_tokens": 336,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ab5e3281-7e77-48fa-be31-8c318924d88a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009635999999999999,
      "cost_per_success_usd": 0.0032119999999999996,
      "octane_score": 1.22156496471565,
      "ci_low": 1.1636022143139584,
      "ci_high": 1.3007968127490044,
      "ci_metric": "octane_score",
      "median_latency_ms": 3442.702959291637,
      "p95_latency_ms": 3722.74515805766,
      "input_tokens": 267,
      "output_tokens": 589,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "07c82a67-2a57-4614-8d95-32d56f152d68",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00012198,
      "cost_per_success_usd": 4.066e-05,
      "octane_score": 51.5494343334973,
      "ci_low": 51.5494343334973,
      "ci_high": 51.5494343334973,
      "ci_metric": "octane_score",
      "median_latency_ms": 816.0546664148569,
      "p95_latency_ms": 1106.5401658415794,
      "input_tokens": 456,
      "output_tokens": 9,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3a33ba12-9744-4aad-8a34-7ba19c935a31",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 7.404e-05,
      "cost_per_success_usd": 2.468e-05,
      "octane_score": 71.54240950837385,
      "ci_low": 70.62398703403566,
      "ci_high": 72.00162074554295,
      "ci_metric": "octane_score",
      "median_latency_ms": 1222.1194170415401,
      "p95_latency_ms": 1980.4921920411289,
      "input_tokens": 276,
      "output_tokens": 6,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "176aa4f2-6674-4c0c-aa38-a51fd9adb660",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00011806,
      "cost_per_success_usd": 3.9353333333333335e-05,
      "octane_score": 59.86786379806877,
      "ci_low": 54.388915585455706,
      "ci_high": 67.66226306720276,
      "ci_metric": "octane_score",
      "median_latency_ms": 2141.0040836781263,
      "p95_latency_ms": 2185.658633429557,
      "input_tokens": 318,
      "output_tokens": 115,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d1c68184-5a74-42a2-a3cf-58864cbbdece",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00024735999999999996,
      "cost_per_success_usd": 8.245333333333333e-05,
      "octane_score": 47.58651358344115,
      "ci_low": 40.02988318391742,
      "ci_high": 53.05183199285076,
      "ci_metric": "octane_score",
      "median_latency_ms": 3477.7755830436945,
      "p95_latency_ms": 4352.231184113771,
      "input_tokens": 225,
      "output_tokens": 497,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e025cb81-87d5-4298-bdb8-d5b0d91bdcc4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 4.5480000000000005e-05,
      "cost_per_success_usd": 1.5160000000000002e-05,
      "octane_score": 116.31486367634123,
      "ci_low": 79.68337730870711,
      "ci_high": 140.2374670184697,
      "ci_metric": "octane_score",
      "median_latency_ms": 1388.786000199616,
      "p95_latency_ms": 1618.2199624367058,
      "input_tokens": 153,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c4922768-31f4-4b7f-8ac0-732069dd5d4f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00013098,
      "cost_per_success_usd": 4.366e-05,
      "octane_score": 155.5581004733547,
      "ci_low": 152.01557489693084,
      "ci_high": 162.13925790196978,
      "ci_metric": "octane_score",
      "median_latency_ms": 1147.6237503811717,
      "p95_latency_ms": 1189.2365619540215,
      "input_tokens": 495,
      "output_tokens": 6,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "17bbf413-cb8a-4c3f-a329-550ee21728f3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 7.792e-05,
      "cost_per_success_usd": 3.896e-05,
      "octane_score": 55.321697467488036,
      "ci_low": 26.838671411625153,
      "ci_high": 85.49766718506999,
      "ci_metric": "octane_score",
      "median_latency_ms": 1168.9025005325675,
      "p95_latency_ms": 1197.8316502645612,
      "input_tokens": 288,
      "output_tokens": 8,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5af50d2c-6fa0-49bf-af96-37a7bfad0148",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 9.036e-05,
      "cost_per_success_usd": 3.012e-05,
      "octane_score": 61.10004426737495,
      "ci_low": 44.4887118193891,
      "ci_high": 77.78884462151395,
      "ci_metric": "octane_score",
      "median_latency_ms": 1301.7198331654072,
      "p95_latency_ms": 1419.5850701071322,
      "input_tokens": 330,
      "output_tokens": 12,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b9837e3e-4d4f-478b-991e-19949a66f3ff",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 7.578e-05,
      "cost_per_success_usd": 2.526e-05,
      "octane_score": 109.80469780944844,
      "ci_low": 105.10688836104516,
      "ci_high": 114.4893111638955,
      "ci_metric": "octane_score",
      "median_latency_ms": 1292.9327497258782,
      "p95_latency_ms": 2102.187100145966,
      "input_tokens": 195,
      "output_tokens": 66,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3db56add-d905-4e7b-bf48-af54d6fc1821",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 6.944e-05,
      "cost_per_success_usd": 2.3146666666666666e-05,
      "octane_score": 79.50748847926268,
      "ci_low": 58.80631948507899,
      "ci_high": 101.22407834101382,
      "ci_metric": "octane_score",
      "median_latency_ms": 1656.2810000032187,
      "p95_latency_ms": 6409.558662958442,
      "input_tokens": 391,
      "output_tokens": 11,
      "reasoning_tokens": 201,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4dd29c16-cd68-4c00-a0a8-6ec100d179a1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 6.564e-05,
      "cost_per_success_usd": 2.188e-05,
      "octane_score": 80.69774527726996,
      "ci_low": 68.05016808895785,
      "ci_high": 125.4024621212121,
      "ci_metric": "octane_score",
      "median_latency_ms": 1032.9310838133097,
      "p95_latency_ms": 3628.7626092322166,
      "input_tokens": 258,
      "output_tokens": 34,
      "reasoning_tokens": 216,
      "undefined_reason": null
    },
    {
      "snapshot_id": "17990455-a9d1-4e8c-9052-dd053137b4ee",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 7.214e-05,
      "cost_per_success_usd": 2.4046666666666665e-05,
      "octane_score": 115.34516218464101,
      "ci_low": 98.00942285041228,
      "ci_high": 145.75337597725664,
      "ci_metric": "octane_score",
      "median_latency_ms": 4520.492499694228,
      "p95_latency_ms": 12164.570837654172,
      "input_tokens": 256,
      "output_tokens": 64,
      "reasoning_tokens": 223,
      "undefined_reason": null
    },
    {
      "snapshot_id": "461afa51-8ecc-44e4-94f8-1dfd2d3dfb93",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00013126000000000001,
      "cost_per_success_usd": 4.375333333333334e-05,
      "octane_score": 155.22626847478284,
      "ci_low": 102.70392749244712,
      "ci_high": 240.83924349881798,
      "ci_metric": "octane_score",
      "median_latency_ms": 3847.0359575003386,
      "p95_latency_ms": 6615.01865806058,
      "input_tokens": 635,
      "output_tokens": 295,
      "reasoning_tokens": 152,
      "undefined_reason": null
    },
    {
      "snapshot_id": "934741b1-265a-42de-b774-f014d51285fd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 8.014e-05,
      "cost_per_success_usd": 2.6713333333333334e-05,
      "octane_score": 80.68380334414776,
      "ci_low": 71.78894029521922,
      "ci_high": 94.28277483833038,
      "ci_metric": "octane_score",
      "median_latency_ms": 2321.620042435825,
      "p95_latency_ms": 2628.46986670047,
      "input_tokens": 428,
      "output_tokens": 113,
      "reasoning_tokens": 142,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a209aef7-abdb-4482-9a81-fb0079e33ab1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 7.838e-05,
      "cost_per_success_usd": 2.612666666666667e-05,
      "octane_score": 90.1760653227864,
      "ci_low": 79.0784268094019,
      "ci_high": 111.36865342163357,
      "ci_metric": "octane_score",
      "median_latency_ms": 6971.87350038439,
      "p95_latency_ms": 7125.805187411606,
      "input_tokens": 379,
      "output_tokens": 93,
      "reasoning_tokens": 174,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0a5341ed-b6d0-4a37-8d4a-56b75b8aa2a6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 5.078e-05,
      "cost_per_success_usd": 1.6926666666666665e-05,
      "octane_score": 104.17487199684916,
      "ci_low": 71.36667979519495,
      "ci_high": 125.41880363302185,
      "ci_metric": "octane_score",
      "median_latency_ms": 2459.62275005877,
      "p95_latency_ms": 6498.59643774107,
      "input_tokens": 214,
      "output_tokens": 22,
      "reasoning_tokens": 165,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3f7e538c-6f60-4837-99f4-ad3e5c1e0c7f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 6.71e-05,
      "cost_per_success_usd": 2.2366666666666667e-05,
      "octane_score": 93.71087928464978,
      "ci_low": 67.70025839793281,
      "ci_high": 116.57397107897665,
      "ci_metric": "octane_score",
      "median_latency_ms": 1079.9365835264325,
      "p95_latency_ms": 1574.8046580702064,
      "input_tokens": 517,
      "output_tokens": 73,
      "reasoning_tokens": 70,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e96b89b8-c80d-476d-ae03-fc820f990808",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00010298,
      "cost_per_success_usd": 3.432666666666667e-05,
      "octane_score": 114.3037483006409,
      "ci_low": 100.28926322953888,
      "ci_high": 129.9227373068433,
      "ci_metric": "octane_score",
      "median_latency_ms": 2196.6068744659424,
      "p95_latency_ms": 2450.1563373953104,
      "input_tokens": 286,
      "output_tokens": 121,
      "reasoning_tokens": 324,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c6c7562c-3297-45af-b93a-465c8971346f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00028971,
      "cost_per_success_usd": 9.656999999999999e-05,
      "octane_score": 28.72182527354942,
      "ci_low": 20.051085568326954,
      "ci_high": 52.841044463762415,
      "ci_metric": "octane_score",
      "median_latency_ms": 1990.416250191629,
      "p95_latency_ms": 2262.927063368261,
      "input_tokens": 256,
      "output_tokens": 75,
      "reasoning_tokens": 130,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f33dc2fe-5650-499a-93ce-954b3bfbe75c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005311350000000001,
      "cost_per_success_usd": 0.00017704500000000002,
      "octane_score": 11.838798045694597,
      "ci_low": 11.127923336253353,
      "ci_high": 12.450622234102587,
      "ci_metric": "octane_score",
      "median_latency_ms": 2307.140249758959,
      "p95_latency_ms": 2555.9146505780514,
      "input_tokens": 675,
      "output_tokens": 9,
      "reasoning_tokens": 264,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ebcf831a-e1c3-490b-8a00-f8b3c42e4782",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000478065,
      "cost_per_success_usd": 0.000159355,
      "octane_score": 11.548638783429032,
      "ci_low": 8.533129557105102,
      "ci_high": 14.703021555646199,
      "ci_metric": "octane_score",
      "median_latency_ms": 2675.7413325831294,
      "p95_latency_ms": 3023.121608234942,
      "input_tokens": 549,
      "output_tokens": 12,
      "reasoning_tokens": 263,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8ada2157-c1dd-4cbd-aa7a-d750723371e2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00038758499999999997,
      "cost_per_success_usd": 0.000129195,
      "octane_score": 18.235999845195245,
      "ci_low": 16.14906832298137,
      "ci_high": 20.692472391255354,
      "ci_metric": "octane_score",
      "median_latency_ms": 2997.2516670823097,
      "p95_latency_ms": 4041.4642171934247,
      "input_tokens": 379,
      "output_tokens": 95,
      "reasoning_tokens": 161,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5bfca323-e8e6-4622-9b07-e9d9f0e3ce3f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00026882999999999996,
      "cost_per_success_usd": 8.960999999999999e-05,
      "octane_score": 19.677863333705318,
      "ci_low": 13.740908071389676,
      "ci_high": 23.54994120101666,
      "ci_metric": "octane_score",
      "median_latency_ms": 1824.8023334890604,
      "p95_latency_ms": 4284.38814599067,
      "input_tokens": 372,
      "output_tokens": 13,
      "reasoning_tokens": 110,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ff8a2bab-2c8d-4465-b8e6-693137b84f86",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00054288,
      "cost_per_success_usd": 0.00018096,
      "octane_score": 37.53131447096964,
      "ci_low": 33.07844665238002,
      "ci_high": 41.290053851417696,
      "ci_metric": "octane_score",
      "median_latency_ms": 1780.1354173570871,
      "p95_latency_ms": 2727.4805915541942,
      "input_tokens": 556,
      "output_tokens": 47,
      "reasoning_tokens": 299,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6e3461e6-1c9b-4128-8cb6-941d51a6c68c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00032364,
      "cost_per_success_usd": 0.00010787999999999999,
      "octane_score": 16.366950933135584,
      "ci_low": 13.75933502175466,
      "ci_high": 25.36877394636016,
      "ci_metric": "octane_score",
      "median_latency_ms": 2076.457583345473,
      "p95_latency_ms": 2701.717883814126,
      "input_tokens": 416,
      "output_tokens": 11,
      "reasoning_tokens": 153,
      "undefined_reason": null
    },
    {
      "snapshot_id": "06ce6c9e-d9c2-4930-abed-55a5b096e4b5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000640755,
      "cost_per_success_usd": 0.000213585,
      "octane_score": 18.370516031868657,
      "ci_low": 14.619000602346047,
      "ci_high": 21.290368856133554,
      "ci_metric": "octane_score",
      "median_latency_ms": 3973.493916913867,
      "p95_latency_ms": 4158.003254421055,
      "input_tokens": 365,
      "output_tokens": 282,
      "reasoning_tokens": 272,
      "undefined_reason": null
    },
    {
      "snapshot_id": "504885a0-3f4c-45fa-a4c0-338eabf67483",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.00032886,
      "cost_per_success_usd": 0.00016443,
      "octane_score": 13.107908127065219,
      "ci_low": 9.389566983110639,
      "ci_high": 15.303246459514945,
      "ci_metric": "octane_score",
      "median_latency_ms": 2576.264957897365,
      "p95_latency_ms": 4340.2356329374015,
      "input_tokens": 428,
      "output_tokens": 5,
      "reasoning_tokens": 159,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b5b5a256-cb9a-49e1-aa8f-071b529dc705",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0001802,
      "cost_per_success_usd": 9.01e-05,
      "octane_score": 75.37920828708842,
      "ci_low": 37.39073344588258,
      "ci_high": 113.94088669950739,
      "ci_metric": "octane_score",
      "median_latency_ms": 596.9058750197291,
      "p95_latency_ms": 659.5030494034289,
      "input_tokens": 534,
      "output_tokens": 8,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ef169583-b181-4bc3-9b87-7542889d548b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001491,
      "cost_per_success_usd": 4.9699999999999995e-05,
      "octane_score": 42.17303822937627,
      "ci_low": 42.17303822937627,
      "ci_high": 42.17303822937627,
      "ci_metric": "octane_score",
      "median_latency_ms": 627.4288333952427,
      "p95_latency_ms": 696.1484834551811,
      "input_tokens": 447,
      "output_tokens": 6,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "84cc547e-6d5e-4ee6-bd79-1e975b5a7ea4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 7.71e-05,
      "cost_per_success_usd": 2.57e-05,
      "octane_score": 68.61219195849546,
      "ci_low": 47.00389105058365,
      "ci_high": 82.72373540856033,
      "ci_metric": "octane_score",
      "median_latency_ms": 589.973459020257,
      "p95_latency_ms": 647.4685337394475,
      "input_tokens": 132,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b11cd4ed-1dbc-43ab-974c-cd01813d5892",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016349999999999997,
      "cost_per_success_usd": 0.0005449999999999999,
      "octane_score": 7.199388379204895,
      "ci_low": 7.1266055045871575,
      "ci_high": 7.313761467889909,
      "ci_metric": "octane_score",
      "median_latency_ms": 1396.8349993228912,
      "p95_latency_ms": 1412.8520384430885,
      "input_tokens": 225,
      "output_tokens": 627,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c7bfa8ba-447e-43b2-820a-00f10457d3f6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0004344,
      "cost_per_success_usd": 0.0001448,
      "octane_score": 16.270718232044196,
      "ci_low": 15.62154696132597,
      "ci_high": 17.534530386740332,
      "ci_metric": "octane_score",
      "median_latency_ms": 876.8744161352515,
      "p95_latency_ms": 1247.7316414937375,
      "input_tokens": 273,
      "output_tokens": 141,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5f274a6d-5b1e-4e59-96fd-2ace7945ccaf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 9.69e-05,
      "cost_per_success_usd": 4.845e-05,
      "octane_score": 44.485724114207095,
      "ci_low": 21.792225662194706,
      "ci_high": 68.08049535603715,
      "ci_metric": "octane_score",
      "median_latency_ms": 650.6385831162333,
      "p95_latency_ms": 687.1712321415544,
      "input_tokens": 273,
      "output_tokens": 6,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fa8d6f72-7a88-4b2a-aa6d-9d63a0553924",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000117,
      "cost_per_success_usd": 3.9e-05,
      "octane_score": 47.18803418803419,
      "ci_low": 34.35897435897436,
      "ci_high": 60.07692307692308,
      "ci_metric": "octane_score",
      "median_latency_ms": 653.757625259459,
      "p95_latency_ms": 748.7429495900869,
      "input_tokens": 315,
      "output_tokens": 9,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f6cbff5e-0154-410a-99a5-4beeab448cb8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 9.12e-05,
      "cost_per_success_usd": 3.0399999999999997e-05,
      "octane_score": 58.0811403508772,
      "ci_low": 57.33552631578949,
      "ci_high": 58.45394736842106,
      "ci_metric": "octane_score",
      "median_latency_ms": 526.1428337544203,
      "p95_latency_ms": 665.0078838691114,
      "input_tokens": 279,
      "output_tokens": 3,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bf9cccfd-33d1-418c-8e0e-f1062719e9af",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0003099,
      "cost_per_success_usd": 0.0001033,
      "octane_score": 26.850596966763476,
      "ci_low": 25.701839303000973,
      "ci_high": 27.996127783155856,
      "ci_metric": "octane_score",
      "median_latency_ms": 633.8185407221317,
      "p95_latency_ms": 665.0965788401663,
      "input_tokens": 183,
      "output_tokens": 102,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "acda17eb-4ae7-4dc3-bdd8-448436d1e756",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0092505,
      "cost_per_success_usd": 0.0030835,
      "octane_score": 0.6797470407005027,
      "ci_low": 0.636888483743543,
      "ci_high": 0.7078689631881122,
      "ci_metric": "octane_score",
      "median_latency_ms": 2700.751166790724,
      "p95_latency_ms": 3195.3384794294834,
      "input_tokens": 447,
      "output_tokens": 6,
      "reasoning_tokens": 1138,
      "undefined_reason": null
    },
    {
      "snapshot_id": "22ba98b5-fe4a-4e17-a040-8227cd743980",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.011145,
      "cost_per_success_usd": 0.003715,
      "octane_score": 1.8281740690892776,
      "ci_low": 1.7519030993065017,
      "ci_high": 1.9078911190541654,
      "ci_metric": "octane_score",
      "median_latency_ms": 2979.463666677475,
      "p95_latency_ms": 3339.494791533798,
      "input_tokens": 535,
      "output_tokens": 9,
      "reasoning_tokens": 1370,
      "undefined_reason": null
    },
    {
      "snapshot_id": "034012db-9958-47c0-856e-5eacc2d06a29",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.006295500000000001,
      "cost_per_success_usd": 0.0020985,
      "octane_score": 0.841394646970058,
      "ci_low": 0.7384636832566568,
      "ci_high": 0.955519244734931,
      "ci_metric": "octane_score",
      "median_latency_ms": 1682.1831250563264,
      "p95_latency_ms": 1863.0647120065987,
      "input_tokens": 282,
      "output_tokens": 3,
      "reasoning_tokens": 780,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c2d22249-e1aa-4664-a538-8ffd68c5130e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.011247,
      "cost_per_success_usd": 0.003749,
      "octane_score": 0.739841735573931,
      "ci_low": 0.5904701931150294,
      "ci_high": 0.9104436785382137,
      "ci_metric": "octane_score",
      "median_latency_ms": 2950.5581660196185,
      "p95_latency_ms": 3551.9657669588923,
      "input_tokens": 183,
      "output_tokens": 69,
      "reasoning_tokens": 1394,
      "undefined_reason": null
    },
    {
      "snapshot_id": "34a7a12b-24e4-4e3b-8701-72b826e0f120",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007261500000000001,
      "cost_per_success_usd": 0.0024205000000000003,
      "octane_score": 0.7603112304620256,
      "ci_low": 0.560083594566353,
      "ci_high": 0.9569133111663901,
      "ci_metric": "octane_score",
      "median_latency_ms": 2619.300624355674,
      "p95_latency_ms": 2768.2721491903067,
      "input_tokens": 316,
      "output_tokens": 9,
      "reasoning_tokens": 896,
      "undefined_reason": null
    },
    {
      "snapshot_id": "95f4cf1e-133e-4f73-b24d-a4535a96f370",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009546,
      "cost_per_success_usd": 0.0031820000000000004,
      "octane_score": 0.6773517703750263,
      "ci_low": 0.5635434155581386,
      "ci_high": 0.8368601566038958,
      "ci_metric": "octane_score",
      "median_latency_ms": 2945.659415796399,
      "p95_latency_ms": 3493.1378550827503,
      "input_tokens": 274,
      "output_tokens": 6,
      "reasoning_tokens": 1212,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7d3b5495-e78c-40b7-b9aa-789a7d80623c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0101775,
      "cost_per_success_usd": 0.0033925,
      "octane_score": 0.6944731024318348,
      "ci_low": 0.6326625675423887,
      "ci_high": 0.7634341544537989,
      "ci_metric": "octane_score",
      "median_latency_ms": 2784.9466670304537,
      "p95_latency_ms": 2985.799329355359,
      "input_tokens": 275,
      "output_tokens": 141,
      "reasoning_tokens": 1161,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d2eb5bc7-9e63-45d5-a69c-561a96f621a0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.013449,
      "cost_per_success_usd": 0.004483,
      "octane_score": 0.8752323592832182,
      "ci_low": 0.6855680595539272,
      "ci_high": 1.5523903725684145,
      "ci_metric": "octane_score",
      "median_latency_ms": 3913.240084424615,
      "p95_latency_ms": 4280.455496069044,
      "input_tokens": 226,
      "output_tokens": 12,
      "reasoning_tokens": 1736,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a0b65111-d25c-44ca-8d50-7f5f2007ea5e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002901,
      "cost_per_success_usd": 0.000967,
      "octane_score": 1.82350913478111,
      "ci_low": 1.2589890568004167,
      "ci_high": 2.192090395480226,
      "ci_metric": "octane_score",
      "median_latency_ms": 1356.3624154776335,
      "p95_latency_ms": 1490.0046170689166,
      "input_tokens": 134,
      "output_tokens": 15,
      "reasoning_tokens": 345,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8e640b39-c26d-4c81-b61d-e85b1fa5543f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002289,
      "cost_per_success_usd": 0.0007629999999999999,
      "octane_score": 8.901266928789866,
      "ci_low": 7.406318082788671,
      "ci_high": 10.511381956535757,
      "ci_metric": "octane_score",
      "median_latency_ms": 3234.985792078078,
      "p95_latency_ms": 3573.9993299357593,
      "input_tokens": 534,
      "output_tokens": 9,
      "reasoning_tokens": 1105,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a56c4b1a-0231-4ad9-9d43-008a81535ca9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016395,
      "cost_per_success_usd": 0.0005465,
      "octane_score": 3.8353156450137242,
      "ci_low": 3.441707717569787,
      "ci_high": 4.237553702299722,
      "ci_metric": "octane_score",
      "median_latency_ms": 2834.389832802117,
      "p95_latency_ms": 3282.62032084167,
      "input_tokens": 447,
      "output_tokens": 6,
      "reasoning_tokens": 779,
      "undefined_reason": null
    },
    {
      "snapshot_id": "37f80044-370f-4451-9b1d-b05da531d934",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00205875,
      "cost_per_success_usd": 0.0006862499999999999,
      "octane_score": 5.717547055251975,
      "ci_low": 4.762828282828283,
      "ci_high": 7.059459459459461,
      "ci_metric": "octane_score",
      "median_latency_ms": 2909.6420845016837,
      "p95_latency_ms": 3225.235783588141,
      "input_tokens": 225,
      "output_tokens": 12,
      "reasoning_tokens": 1041,
      "undefined_reason": null
    },
    {
      "snapshot_id": "53a4bd16-03a5-446f-a851-07f603d24d0a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002221125,
      "cost_per_success_usd": 0.000740375,
      "octane_score": 3.182171197028532,
      "ci_low": 2.8891725164858544,
      "ci_high": 3.5114138878929126,
      "ci_metric": "octane_score",
      "median_latency_ms": 3282.19879232347,
      "p95_latency_ms": 3393.0381797254086,
      "input_tokens": 273,
      "output_tokens": 141,
      "reasoning_tokens": 989,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7f746e37-1950-48bd-bcc4-62ed549d0c84",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001008,
      "cost_per_success_usd": 0.000336,
      "octane_score": 6.414682539682541,
      "ci_low": 6.147501213003397,
      "ci_high": 6.69406392694064,
      "ci_metric": "octane_score",
      "median_latency_ms": 2366.45045876503,
      "p95_latency_ms": 2396.901883650571,
      "input_tokens": 273,
      "output_tokens": 6,
      "reasoning_tokens": 477,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1ebe165b-9a30-4fd2-a554-898e5de2b5fc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00121125,
      "cost_per_success_usd": 0.00040375000000000003,
      "octane_score": 4.558101135190919,
      "ci_low": 3.2213903743315506,
      "ci_high": 6.402346683354159,
      "ci_metric": "octane_score",
      "median_latency_ms": 2815.340249799192,
      "p95_latency_ms": 3186.8721747770905,
      "input_tokens": 315,
      "output_tokens": 9,
      "reasoning_tokens": 574,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ee6f831c-edb6-4cb8-9045-5dd22f09d4fe",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001208625,
      "cost_per_success_usd": 0.000402875,
      "octane_score": 6.884683007549903,
      "ci_low": 6.185061315496099,
      "ci_high": 7.860196008973905,
      "ci_metric": "octane_score",
      "median_latency_ms": 2328.5978753119707,
      "p95_latency_ms": 2423.3916505239904,
      "input_tokens": 183,
      "output_tokens": 84,
      "reasoning_tokens": 524,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e83a611c-3199-434e-ae97-e4017ad68707",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00110025,
      "cost_per_success_usd": 0.00036675000000000005,
      "octane_score": 4.8143603726425805,
      "ci_low": 4.014019134223737,
      "ci_high": 7.438300860101807,
      "ci_metric": "octane_score",
      "median_latency_ms": 2611.812250688672,
      "p95_latency_ms": 2618.3392747305334,
      "input_tokens": 279,
      "output_tokens": 3,
      "reasoning_tokens": 528,
      "undefined_reason": null
    },
    {
      "snapshot_id": "afe7000c-3455-4994-9f99-3743555f88e3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00069825,
      "cost_per_success_usd": 0.00023275,
      "octane_score": 7.57608306480487,
      "ci_low": 5.318657127132636,
      "ci_high": 9.449010654490108,
      "ci_metric": "octane_score",
      "median_latency_ms": 2129.3726256117225,
      "p95_latency_ms": 2136.4316632971168,
      "input_tokens": 132,
      "output_tokens": 15,
      "reasoning_tokens": 331,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ae189141-a283-4346-b8e3-a74018c1bafe",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0006103,
      "cost_per_success_usd": 0.00020343333333333334,
      "octane_score": 11.581189578895623,
      "ci_low": 10.6995430912242,
      "ci_high": 12.536268987881888,
      "ci_metric": "octane_score",
      "median_latency_ms": 6604.6844171360135,
      "p95_latency_ms": 7143.766378797591,
      "input_tokens": 285,
      "output_tokens": 225,
      "reasoning_tokens": 1274,
      "undefined_reason": null
    },
    {
      "snapshot_id": "14c57897-88dd-4d18-acb1-15825a54112d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": 1,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.0068324200000000005,
      "cost_per_success_usd": 0.0068324200000000005,
      "octane_score": 0.25808327552072813,
      "ci_low": 0.10996401772100829,
      "ci_high": 12.618672773245553,
      "ci_metric": "octane_score",
      "median_latency_ms": 7011.980916373432,
      "p95_latency_ms": 202258.8064042851,
      "input_tokens": 147,
      "output_tokens": 1753,
      "reasoning_tokens": 15306,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d15efea9-6f02-4d0a-9910-4db9189a2911",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0010136200000000002,
      "cost_per_success_usd": 0.0003378733333333334,
      "octane_score": 11.612833211657227,
      "ci_low": 10.077738394890499,
      "ci_high": 13.945091871023093,
      "ci_metric": "octane_score",
      "median_latency_ms": 12727.796416729689,
      "p95_latency_ms": 13056.067328620702,
      "input_tokens": 207,
      "output_tokens": 722,
      "reasoning_tokens": 1797,
      "undefined_reason": null
    },
    {
      "snapshot_id": "986b926b-5781-47bf-bd9e-d54e674f926c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00059332,
      "cost_per_success_usd": 0.00019777333333333334,
      "octane_score": 10.597990966089126,
      "ci_low": 10.33326760007888,
      "ci_high": 11.122903842071748,
      "ci_metric": "octane_score",
      "median_latency_ms": 8245.623541995883,
      "p95_latency_ms": 41830.307792127125,
      "input_tokens": 462,
      "output_tokens": 208,
      "reasoning_tokens": 1206,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cadb4455-b0af-4008-9df2-e5f86b9649a3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0006240200000000001,
      "cost_per_success_usd": 0.0002080066666666667,
      "octane_score": 10.361847376686645,
      "ci_low": 10.03425410892178,
      "ci_high": 10.695178496157713,
      "ci_metric": "octane_score",
      "median_latency_ms": 7920.521167106926,
      "p95_latency_ms": 24180.306779593226,
      "input_tokens": 267,
      "output_tokens": 381,
      "reasoning_tokens": 1147,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ae05ee55-8a4a-4f9f-b26c-9e7ce7a790ed",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005815200000000001,
      "cost_per_success_usd": 0.00019384000000000003,
      "octane_score": 9.494084468289998,
      "ci_low": 6.960797894445212,
      "ci_high": 12.08728848534874,
      "ci_metric": "octane_score",
      "median_latency_ms": 5989.5089995115995,
      "p95_latency_ms": 6546.699674893171,
      "input_tokens": 312,
      "output_tokens": 44,
      "reasoning_tokens": 1379,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b7b6c6bc-5fe8-47d3-a5bd-ee1302f3b227",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00068386,
      "cost_per_success_usd": 0.00022795333333333332,
      "octane_score": 12.167695142280587,
      "ci_low": 11.268650564669029,
      "ci_high": 13.408339257564988,
      "ci_metric": "octane_score",
      "median_latency_ms": 6402.135208249092,
      "p95_latency_ms": 7509.451783075929,
      "input_tokens": 171,
      "output_tokens": 626,
      "reasoning_tokens": 1058,
      "undefined_reason": null
    },
    {
      "snapshot_id": "294a2b81-f2d7-41ff-9cc4-29328476e0e8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00043638,
      "cost_per_success_usd": 0.00014545999999999998,
      "octane_score": 12.138503139465605,
      "ci_low": 11.045081112640228,
      "ci_high": 12.996221600667353,
      "ci_metric": "octane_score",
      "median_latency_ms": 13877.198500558734,
      "p95_latency_ms": 22061.149037629362,
      "input_tokens": 273,
      "output_tokens": 226,
      "reasoning_tokens": 824,
      "undefined_reason": null
    },
    {
      "snapshot_id": "36323338-c0d1-4a52-9465-f8d6becae208",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00103174,
      "cost_per_success_usd": 0.0003439133333333333,
      "octane_score": 19.748192374047726,
      "ci_low": 18.506723283793345,
      "ci_high": 21.015809558422678,
      "ci_metric": "octane_score",
      "median_latency_ms": 8574.065083637834,
      "p95_latency_ms": 8722.158807795495,
      "input_tokens": 489,
      "output_tokens": 703,
      "reasoning_tokens": 1835,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6018e967-4fe1-4c40-b6cc-0c87ebe3100c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002595,
      "cost_per_success_usd": 0.000865,
      "octane_score": 2.423121387283237,
      "ci_low": 1.828204591445119,
      "ci_high": 5.766479586222076,
      "ci_metric": "octane_score",
      "median_latency_ms": 3541.193500161171,
      "p95_latency_ms": 8292.928937356919,
      "input_tokens": 469,
      "output_tokens": 60,
      "reasoning_tokens": 1145,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b67bf552-20dd-4788-9075-f2e59e7a6ea8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0035849999999999996,
      "cost_per_success_usd": 0.0011949999999999999,
      "octane_score": 5.683403068340307,
      "ci_low": 2.8948505049471756,
      "ci_high": 15.354728710242018,
      "ci_metric": "octane_score",
      "median_latency_ms": 4555.182874202728,
      "p95_latency_ms": 12575.159187521784,
      "input_tokens": 503,
      "output_tokens": 9,
      "reasoning_tokens": 1701,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e8faf26d-8797-4c16-b386-497b0041b797",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0023821199999999997,
      "cost_per_success_usd": 0.0007940399999999999,
      "octane_score": 2.223649522274277,
      "ci_low": 2.0362425808038873,
      "ci_high": 2.445541057953236,
      "ci_metric": "octane_score",
      "median_latency_ms": 6796.718209050596,
      "p95_latency_ms": 11367.948146071283,
      "input_tokens": 287,
      "output_tokens": 71,
      "reasoning_tokens": 1080,
      "undefined_reason": null
    },
    {
      "snapshot_id": "83fc2612-691c-4171-8c4e-372f22015f1a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0036184799999999994,
      "cost_per_success_usd": 0.0012061599999999997,
      "octane_score": 2.299584355862131,
      "ci_low": 2.063487201478442,
      "ci_high": 2.7074751795521768,
      "ci_metric": "octane_score",
      "median_latency_ms": 9052.65829153359,
      "p95_latency_ms": 10584.199766442178,
      "input_tokens": 178,
      "output_tokens": 74,
      "reasoning_tokens": 1755,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d2597be9-ce9c-4beb-96d7-04b045d82840",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003048,
      "cost_per_success_usd": 0.001016,
      "octane_score": 1.8113517060367454,
      "ci_low": 1.2952387505160314,
      "ci_high": 2.5016872200748517,
      "ci_metric": "octane_score",
      "median_latency_ms": 61700.81774983555,
      "p95_latency_ms": 71476.51518778875,
      "input_tokens": 312,
      "output_tokens": 14,
      "reasoning_tokens": 1505,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5976d44a-9303-4e25-a9fd-3eae6f5d8e55",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00251688,
      "cost_per_success_usd": 0.00083896,
      "octane_score": 2.5690537490861707,
      "ci_low": 1.5604497696558728,
      "ci_high": 5.153833891280091,
      "ci_metric": "octane_score",
      "median_latency_ms": 49380.84708433598,
      "p95_latency_ms": 114357.35875852405,
      "input_tokens": 274,
      "output_tokens": 10,
      "reasoning_tokens": 1216,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a5d109da-09d5-4990-bebb-c71729f9d167",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0030798,
      "cost_per_success_usd": 0.0010266000000000001,
      "octane_score": 2.2949542178063504,
      "ci_low": 2.0233711132563434,
      "ci_high": 2.7781708409980657,
      "ci_metric": "octane_score",
      "median_latency_ms": 6703.152917325497,
      "p95_latency_ms": 7779.8188663087785,
      "input_tokens": 285,
      "output_tokens": 126,
      "reasoning_tokens": 1405,
      "undefined_reason": null
    },
    {
      "snapshot_id": "74d24e34-6a06-4104-8785-edfac0a38c03",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00617604,
      "cost_per_success_usd": 0.00205868,
      "octane_score": 1.9059138218016727,
      "ci_low": 1.7849247448593322,
      "ci_high": 2.0212963536185233,
      "ci_metric": "octane_score",
      "median_latency_ms": 14903.201125562191,
      "p95_latency_ms": 31365.255999844518,
      "input_tokens": 207,
      "output_tokens": 333,
      "reasoning_tokens": 2819,
      "undefined_reason": null
    },
    {
      "snapshot_id": "040db564-5ab9-4071-8e66-b943a104bc74",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00326964,
      "cost_per_success_usd": 0.0010898799999999999,
      "octane_score": 1.617915122154121,
      "ci_low": 1.0170374712707848,
      "ci_high": 2.694910110153001,
      "ci_metric": "octane_score",
      "median_latency_ms": 7670.612166635692,
      "p95_latency_ms": 24236.370891518887,
      "input_tokens": 147,
      "output_tokens": 83,
      "reasoning_tokens": 1574,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f0f07816-f61d-40aa-870c-9264aa5e5bf8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0014383464,
      "cost_per_success_usd": 0.0004794488,
      "octane_score": 3.677834491051669,
      "ci_low": 2.520879424225707,
      "ci_high": 4.984776763264805,
      "ci_metric": "octane_score",
      "median_latency_ms": 2014.933000318706,
      "p95_latency_ms": 2373.5349251888692,
      "input_tokens": 168,
      "output_tokens": 20,
      "reasoning_tokens": 425,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4a564b28-809a-4a05-af0c-1abae41e10fe",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0006808368,
      "cost_per_success_usd": 0.0002269456,
      "octane_score": 7.780131743760032,
      "ci_low": 6.448934470936575,
      "ci_high": 8.773097908898906,
      "ci_metric": "octane_score",
      "median_latency_ms": 1264.629083685577,
      "p95_latency_ms": 1489.1341582871974,
      "input_tokens": 294,
      "output_tokens": 23,
      "reasoning_tokens": 152,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5af46528-f17c-44af-9390-f8e92ea53d94",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0040168488,
      "cost_per_success_usd": 0.0013389496,
      "octane_score": 1.7595882623214492,
      "ci_low": 1.3323289970435412,
      "ci_high": 2.2159205003147235,
      "ci_metric": "octane_score",
      "median_latency_ms": 1461.1951662227511,
      "p95_latency_ms": 6641.764065809547,
      "input_tokens": 306,
      "output_tokens": 178,
      "reasoning_tokens": 1064,
      "undefined_reason": null
    },
    {
      "snapshot_id": "60dd8559-a892-4f65-b719-7afd983ba67f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0035260104,
      "cost_per_success_usd": 0.0011753368,
      "octane_score": 3.3383338857990896,
      "ci_low": 3.278355626938244,
      "ci_high": 3.399459155869025,
      "ci_metric": "octane_score",
      "median_latency_ms": 4708.157041110098,
      "p95_latency_ms": 4847.620479483157,
      "input_tokens": 228,
      "output_tokens": 17,
      "reasoning_tokens": 1122,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3eb26019-e33a-4ad7-9691-832930444a44",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0018655944,
      "cost_per_success_usd": 0.0006218648,
      "octane_score": 3.4659194946125487,
      "ci_low": 3.350788109594838,
      "ci_high": 3.5841791496286404,
      "ci_metric": "octane_score",
      "median_latency_ms": 2786.0177075490355,
      "p95_latency_ms": 2965.0729331187904,
      "input_tokens": 288,
      "output_tokens": 10,
      "reasoning_tokens": 577,
      "undefined_reason": null
    },
    {
      "snapshot_id": "07a766cb-02b9-4e0c-aebc-bfddee242dd4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0015895668,
      "cost_per_success_usd": 0.0005298556,
      "octane_score": 3.4732733471786155,
      "ci_low": 2.4668191529997894,
      "ci_high": 4.841108640592355,
      "ci_metric": "octane_score",
      "median_latency_ms": 2332.7187914401293,
      "p95_latency_ms": 2822.3444038070734,
      "input_tokens": 333,
      "output_tokens": 15,
      "reasoning_tokens": 452,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a2206328-28f8-46ba-a78b-ce6d60344382",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0018992663999999999,
      "cost_per_success_usd": 0.0006330887999999999,
      "octane_score": 4.38116527518204,
      "ci_low": 4.158364341027852,
      "ci_high": 4.614722939756458,
      "ci_metric": "octane_score",
      "median_latency_ms": 3046.8013752251863,
      "p95_latency_ms": 5263.8411251828065,
      "input_tokens": 192,
      "output_tokens": 66,
      "reasoning_tokens": 541,
      "undefined_reason": null
    },
    {
      "snapshot_id": "572efe90-c6ea-4dbe-9988-5171e104b698",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0018232283999999998,
      "cost_per_success_usd": 0.0006077427999999999,
      "octane_score": 3.448827365786975,
      "ci_low": 2.1253821327560734,
      "ci_high": 5.110156045051293,
      "ci_metric": "octane_score",
      "median_latency_ms": 1647.9588327929378,
      "p95_latency_ms": 3113.987245224416,
      "input_tokens": 483,
      "output_tokens": 14,
      "reasoning_tokens": 539,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f66a8f0c-4a75-4be8-b893-1713b1a5dcc6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0017942208,
      "cost_per_success_usd": 0.0005980736000000001,
      "octane_score": 11.355904468390957,
      "ci_low": 9.989047338723564,
      "ci_high": 14.512554802256023,
      "ci_metric": "octane_score",
      "median_latency_ms": 2389.4594591110945,
      "p95_latency_ms": 2549.8784592375155,
      "input_tokens": 510,
      "output_tokens": 9,
      "reasoning_tokens": 534,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8e6f896a-dd59-4fec-b57c-147b6a91818a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.00219716,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3414.4464591518044,
      "p95_latency_ms": 3483.845345955342,
      "input_tokens": 483,
      "output_tokens": 222,
      "reasoning_tokens": 190,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "e4302724-72c7-4ccd-9928-353d3065b173",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001476,
      "cost_per_success_usd": 0.000492,
      "octane_score": 3.5840108401084008,
      "ci_low": 2.455284552845528,
      "ci_high": 4.321138211382115,
      "ci_metric": "octane_score",
      "median_latency_ms": 2282.7676255255938,
      "p95_latency_ms": 3189.754662662744,
      "input_tokens": 168,
      "output_tokens": 18,
      "reasoning_tokens": 264,
      "undefined_reason": null
    },
    {
      "snapshot_id": "02c7c050-84b3-452b-b457-c3a8801cc4cb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00265928,
      "cost_per_success_usd": 0.0008864266666666667,
      "octane_score": 4.426386089467827,
      "ci_low": 2.965087030888419,
      "ci_high": 5.933177420644438,
      "ci_metric": "octane_score",
      "median_latency_ms": 4389.19962476939,
      "p95_latency_ms": 4603.160899691284,
      "input_tokens": 228,
      "output_tokens": 21,
      "reasoning_tokens": 544,
      "undefined_reason": null
    },
    {
      "snapshot_id": "089749a5-620b-4f86-ab21-18ceb442e26a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00253088,
      "cost_per_success_usd": 0.0008436266666666667,
      "octane_score": 2.792704513844986,
      "ci_low": 2.4804082110064924,
      "ci_high": 3.157890211185251,
      "ci_metric": "octane_score",
      "median_latency_ms": 3595.366875641048,
      "p95_latency_ms": 4613.9303997159,
      "input_tokens": 306,
      "output_tokens": 108,
      "reasoning_tokens": 403,
      "undefined_reason": null
    },
    {
      "snapshot_id": "519dc63d-5ae9-4fe2-b20b-ca562d924790",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00114608,
      "cost_per_success_usd": 0.0003820266666666667,
      "octane_score": 5.641840011168505,
      "ci_low": 5.225303439456214,
      "ci_high": 6.023997370151216,
      "ci_metric": "octane_score",
      "median_latency_ms": 2429.8226665705442,
      "p95_latency_ms": 2447.0241415314376,
      "input_tokens": 288,
      "output_tokens": 9,
      "reasoning_tokens": 193,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b0f3ca3b-1840-41a4-b97c-bb87cec7df51",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0014391999999999999,
      "cost_per_success_usd": 0.0004797333333333333,
      "octane_score": 5.781684269038356,
      "ci_low": 5.5343246247915525,
      "ci_high": 6.0099750623441395,
      "ci_metric": "octane_score",
      "median_latency_ms": 2280.789834447205,
      "p95_latency_ms": 2370.0551711954176,
      "input_tokens": 192,
      "output_tokens": 66,
      "reasoning_tokens": 200,
      "undefined_reason": null
    },
    {
      "snapshot_id": "dce6958c-a7d5-430f-af6f-f487039def10",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00141588,
      "cost_per_success_usd": 0.00047196,
      "octane_score": 3.8993417521258866,
      "ci_low": 2.892815981559739,
      "ci_high": 4.964403763030766,
      "ci_metric": "octane_score",
      "median_latency_ms": 2472.1055841073394,
      "p95_latency_ms": 2673.001145850867,
      "input_tokens": 333,
      "output_tokens": 15,
      "reasoning_tokens": 234,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cd6516e7-57ca-4467-a911-4451f997d5f1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0022261600000000005,
      "cost_per_success_usd": 0.0007420533333333335,
      "octane_score": 9.152531713803139,
      "ci_low": 8.583522805567037,
      "ci_high": 10.00639134709931,
      "ci_metric": "octane_score",
      "median_latency_ms": 2630.696166306734,
      "p95_latency_ms": 3047.7343421429396,
      "input_tokens": 510,
      "output_tokens": 9,
      "reasoning_tokens": 401,
      "undefined_reason": null
    },
    {
      "snapshot_id": "872a03bd-85e3-477a-9c5b-ee22e68a5630",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00095208,
      "cost_per_success_usd": 0.00031736,
      "octane_score": 5.563608100159651,
      "ci_low": 4.824225865209471,
      "ci_high": 6.064018564018564,
      "ci_metric": "octane_score",
      "median_latency_ms": 1493.5640832409263,
      "p95_latency_ms": 1523.4646712429821,
      "input_tokens": 294,
      "output_tokens": 9,
      "reasoning_tokens": 147,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d15a424b-c13e-4918-a3f3-4062d24b679a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0003285,
      "cost_per_success_usd": 0.0001095,
      "octane_score": 16.80669710806697,
      "ci_low": 12.237442922374427,
      "ci_high": 21.397260273972602,
      "ci_metric": "octane_score",
      "median_latency_ms": 605.8770408853889,
      "p95_latency_ms": 607.8433171845973,
      "input_tokens": 312,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "39ca94ae-ec69-4583-a810-516aabf6d63c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00027225000000000003,
      "cost_per_success_usd": 9.075000000000001e-05,
      "octane_score": 19.456382001836545,
      "ci_low": 19.206611570247933,
      "ci_high": 19.581267217630852,
      "ci_metric": "octane_score",
      "median_latency_ms": 629.5194160193205,
      "p95_latency_ms": 685.2068042382598,
      "input_tokens": 273,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c226529f-356c-469a-bc1c-5150f64c06bc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 2,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.0005175000000000001,
      "cost_per_success_usd": 0.00025875000000000003,
      "octane_score": 26.24798711755233,
      "ci_low": 12.839291465378421,
      "ci_high": 40.22608695652173,
      "ci_metric": "octane_score",
      "median_latency_ms": 971.4499171823263,
      "p95_latency_ms": 1828.493704460561,
      "input_tokens": 492,
      "output_tokens": 33,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "13fb1f6f-cc0c-4240-9d39-c3d74462efe8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00047024999999999996,
      "cost_per_success_usd": 0.00015675,
      "octane_score": 17.694843168527385,
      "ci_low": 15.827704356338725,
      "ci_high": 20.367473618870267,
      "ci_metric": "octane_score",
      "median_latency_ms": 696.9554163515568,
      "p95_latency_ms": 806.1634046025574,
      "input_tokens": 177,
      "output_tokens": 75,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f146d09f-c8b2-4c7e-8448-d7119d600e98",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0003375,
      "cost_per_success_usd": 0.00011250000000000001,
      "octane_score": 34.87703703703704,
      "ci_low": 34.52444444444444,
      "ci_high": 35.43111111111111,
      "ci_metric": "octane_score",
      "median_latency_ms": 625.4837075248361,
      "p95_latency_ms": 703.7076202221215,
      "input_tokens": 216,
      "output_tokens": 39,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c78194f6-c0d2-416b-be1c-07e7bf741ae7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": 1,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.00035999999999999997,
      "cost_per_success_usd": 0.00035999999999999997,
      "octane_score": 5.987037037037039,
      "ci_low": 5.722673893405602,
      "ci_high": 18.42165242165243,
      "ci_metric": "octane_score",
      "median_latency_ms": 639.6924583241343,
      "p95_latency_ms": 669.0690960735083,
      "input_tokens": 270,
      "output_tokens": 35,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "203c8244-7df5-4bdf-ae10-bc31627cfcd8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.000576,
      "cost_per_success_usd": 0.000192,
      "octane_score": 12.27083333333333,
      "ci_low": 11.781250000000002,
      "ci_high": 13.223958333333334,
      "ci_metric": "octane_score",
      "median_latency_ms": 787.9345417022705,
      "p95_latency_ms": 854.1333293542266,
      "input_tokens": 282,
      "output_tokens": 81,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0e3e19a5-8b68-44a1-8a02-ee212295f6c7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00021375,
      "cost_per_success_usd": 7.125e-05,
      "octane_score": 24.748538011695906,
      "ci_low": 16.95438596491228,
      "ci_high": 29.83859649122808,
      "ci_metric": "octane_score",
      "median_latency_ms": 623.7919172272086,
      "p95_latency_ms": 766.7209917679429,
      "input_tokens": 141,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b9f3386c-5bfe-4db1-af8f-4e528ed8f6e5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00042300000000000004,
      "cost_per_success_usd": 0.000141,
      "octane_score": 14.865248226950355,
      "ci_low": 14.865248226950355,
      "ci_high": 14.865248226950355,
      "ci_metric": "octane_score",
      "median_latency_ms": 602.4088328704238,
      "p95_latency_ms": 697.3363338038325,
      "input_tokens": 456,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "24b4adde-2c2c-469b-a268-ce2a8afe4b92",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.000255,
      "cost_per_success_usd": 8.5e-05,
      "octane_score": 32.631372549019616,
      "ci_low": 31.235294117647065,
      "ci_high": 34.023529411764706,
      "ci_metric": "octane_score",
      "median_latency_ms": 1944.4898329675198,
      "p95_latency_ms": 2008.918508514762,
      "input_tokens": 177,
      "output_tokens": 66,
      "reasoning_tokens": 117,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c30760dd-59b7-403c-8d6c-f2376257d59d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001452,
      "cost_per_success_usd": 4.8400000000000004e-05,
      "octane_score": 38.02341597796143,
      "ci_low": 27.685950413223136,
      "ci_high": 48.40909090909091,
      "ci_metric": "octane_score",
      "median_latency_ms": 1087.5960001721978,
      "p95_latency_ms": 1229.250411875546,
      "input_tokens": 312,
      "output_tokens": 27,
      "reasoning_tokens": 42,
      "undefined_reason": null
    },
    {
      "snapshot_id": "463c7aa5-8b57-44f5-85b4-290bd274d710",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0002388,
      "cost_per_success_usd": 7.96e-05,
      "octane_score": 26.33165829145729,
      "ci_low": 24.14746543778802,
      "ci_high": 27.578947368421055,
      "ci_metric": "octane_score",
      "median_latency_ms": 2112.8471670672297,
      "p95_latency_ms": 2154.7478299587965,
      "input_tokens": 456,
      "output_tokens": 24,
      "reasoning_tokens": 99,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3a140db5-390a-4d76-8cc5-b0d22d66889e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 5.7e-05,
      "cost_per_success_usd": 1.9e-05,
      "octane_score": 92.80701754385964,
      "ci_low": 63.57894736842104,
      "ci_high": 111.89473684210529,
      "ci_metric": "octane_score",
      "median_latency_ms": 908.0475829541683,
      "p95_latency_ms": 1157.2743835859,
      "input_tokens": 141,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "74cc0221-151c-4d6a-b436-b0fa0802a272",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0003072,
      "cost_per_success_usd": 0.0001024,
      "octane_score": 38.31705729166667,
      "ci_low": 37.11513994910941,
      "ci_high": 39.81814634146341,
      "ci_metric": "octane_score",
      "median_latency_ms": 2307.726708240807,
      "p95_latency_ms": 2358.6298829875886,
      "input_tokens": 216,
      "output_tokens": 27,
      "reasoning_tokens": 193,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1150acfc-8b20-457c-876c-eb865d90ca2c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0002436,
      "cost_per_success_usd": 8.120000000000001e-05,
      "octane_score": 29.014778325123146,
      "ci_low": 26.425233644859816,
      "ci_high": 31.87934027777777,
      "ci_metric": "octane_score",
      "median_latency_ms": 1850.127000361681,
      "p95_latency_ms": 2071.896750666201,
      "input_tokens": 282,
      "output_tokens": 87,
      "reasoning_tokens": 69,
      "undefined_reason": null
    },
    {
      "snapshot_id": "77e0729e-6ccb-4ab6-b6e8-23c7a38cceea",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00023040000000000002,
      "cost_per_success_usd": 7.680000000000001e-05,
      "octane_score": 28.064236111111114,
      "ci_low": 27.072649572649578,
      "ci_high": 29.087605606758828,
      "ci_metric": "octane_score",
      "median_latency_ms": 1922.424583695829,
      "p95_latency_ms": 2054.9637955613434,
      "input_tokens": 270,
      "output_tokens": 21,
      "reasoning_tokens": 126,
      "undefined_reason": null
    },
    {
      "snapshot_id": "03cd57e6-bc20-428d-acdd-23df09c88135",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0003372,
      "cost_per_success_usd": 0.00011240000000000001,
      "octane_score": 60.42408066429418,
      "ci_low": 57.87746806039489,
      "ci_high": 63.20535714285714,
      "ci_metric": "octane_score",
      "median_latency_ms": 2432.083625346422,
      "p95_latency_ms": 2584.002874419093,
      "input_tokens": 492,
      "output_tokens": 21,
      "reasoning_tokens": 178,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1db0ca62-be87-4b00-a084-38da9fad0b97",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001362,
      "cost_per_success_usd": 4.54e-05,
      "octane_score": 38.8913362701909,
      "ci_low": 31.195524146054183,
      "ci_high": 72.96143250688705,
      "ci_metric": "octane_score",
      "median_latency_ms": 957.4922081083059,
      "p95_latency_ms": 1257.2204455733297,
      "input_tokens": 273,
      "output_tokens": 19,
      "reasoning_tokens": 49,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c86e3e86-8192-4f28-a7b1-036ac72da981",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0007602,
      "cost_per_success_usd": 0.00025340000000000003,
      "octane_score": 6.95869508024204,
      "ci_low": 4.353909465020576,
      "ci_high": 11.728826728826732,
      "ci_metric": "octane_score",
      "median_latency_ms": 2069.3802926689386,
      "p95_latency_ms": 2166.828915849328,
      "input_tokens": 5031,
      "output_tokens": 123,
      "reasoning_tokens": 71,
      "undefined_reason": null
    },
    {
      "snapshot_id": "63dc80fc-1a4a-4cf7-a224-ea8836e247df",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0018408,
      "cost_per_success_usd": 0.0006136000000000001,
      "octane_score": 3.8396349413298556,
      "ci_low": 3.645251396648045,
      "ci_high": 4.134281372123318,
      "ci_metric": "octane_score",
      "median_latency_ms": 2469.7767915204167,
      "p95_latency_ms": 3126.401416771114,
      "input_tokens": 5916,
      "output_tokens": 312,
      "reasoning_tokens": 236,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9148622c-69aa-4aa5-a110-131e4230c2ba",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0022816,
      "cost_per_success_usd": 0.0007605333333333334,
      "octane_score": 8.930136746143058,
      "ci_low": 8.677085147135644,
      "ci_high": 9.294091903719915,
      "ci_metric": "octane_score",
      "median_latency_ms": 3709.447874687612,
      "p95_latency_ms": 3827.6403746567667,
      "input_tokens": 6932,
      "output_tokens": 114,
      "reasoning_tokens": 632,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4a2d4cbc-4c86-4e1e-8d6c-f523274f248d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00087972,
      "cost_per_success_usd": 0.00029323999999999995,
      "octane_score": 6.021234029009231,
      "ci_low": 3.8737750475354686,
      "ci_high": 9.103251529524988,
      "ci_metric": "octane_score",
      "median_latency_ms": 2224.342666566372,
      "p95_latency_ms": 2635.398166812956,
      "input_tokens": 5541,
      "output_tokens": 96,
      "reasoning_tokens": 124,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7b344b39-0628-4fb4-bbde-f731d4909fb7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0019228000000000001,
      "cost_per_success_usd": 0.0006409333333333334,
      "octane_score": 4.327543166215936,
      "ci_low": 4.148437500000001,
      "ci_high": 4.505608641462401,
      "ci_metric": "octane_score",
      "median_latency_ms": 3227.4378752335906,
      "p95_latency_ms": 3423.2109382748604,
      "input_tokens": 5612,
      "output_tokens": 249,
      "reasoning_tokens": 418,
      "undefined_reason": null
    },
    {
      "snapshot_id": "96489ad0-9267-4353-995c-620e489168b4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0015699999999999998,
      "cost_per_success_usd": 0.0005233333333333332,
      "octane_score": 3.516560509554141,
      "ci_low": 2.5614884669300366,
      "ci_high": 4.427004884953688,
      "ci_metric": "octane_score",
      "median_latency_ms": 2465.782457962632,
      "p95_latency_ms": 4210.2887210436165,
      "input_tokens": 5810,
      "output_tokens": 132,
      "reasoning_tokens": 208,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b77484a8-70b9-4528-89b8-f86acd32b4fd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00183,
      "cost_per_success_usd": 0.00061,
      "octane_score": 3.533333333333334,
      "ci_low": 3.423394446231371,
      "ci_high": 3.6443486907524028,
      "ci_metric": "octane_score",
      "median_latency_ms": 3145.8116248250008,
      "p95_latency_ms": 3397.5020243786275,
      "input_tokens": 5868,
      "output_tokens": 114,
      "reasoning_tokens": 433,
      "undefined_reason": null
    },
    {
      "snapshot_id": "365cc9c1-3555-4f91-afcb-c96b098ae1a8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0021286,
      "cost_per_success_usd": 0.0007095333333333333,
      "octane_score": 5.5299257728084195,
      "ci_low": 5.4543329905580995,
      "ci_high": 5.616851473024704,
      "ci_metric": "octane_score",
      "median_latency_ms": 3669.354417361319,
      "p95_latency_ms": 3782.777667231858,
      "input_tokens": 5891,
      "output_tokens": 132,
      "reasoning_tokens": 660,
      "undefined_reason": null
    },
    {
      "snapshot_id": "61b3bb31-c81d-41c0-8cbe-e9033345380b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0019242,
      "cost_per_success_usd": 0.0006414,
      "octane_score": 3.267851574680387,
      "ci_low": 3.127424649358401,
      "ci_high": 3.404808317089019,
      "ci_metric": "octane_score",
      "median_latency_ms": 3434.995625168085,
      "p95_latency_ms": 3525.251862127334,
      "input_tokens": 6543,
      "output_tokens": 121,
      "reasoning_tokens": 392,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4c47d5a6-98e7-4161-9e7f-0aa023ea81de",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000522,
      "cost_per_success_usd": 0.000174,
      "octane_score": 10.134099616858238,
      "ci_low": 6.942528735632183,
      "ci_high": 12.218390804597705,
      "ci_metric": "octane_score",
      "median_latency_ms": 1159.589291550219,
      "p95_latency_ms": 1556.7055175080893,
      "input_tokens": 141,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7983a004-8d81-4146-b2fa-6b839c357507",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0025340000000000002,
      "cost_per_success_usd": 0.0008446666666666667,
      "octane_score": 8.040647198105761,
      "ci_low": 7.604801829268292,
      "ci_high": 8.51759410801964,
      "ci_metric": "octane_score",
      "median_latency_ms": 4151.401250623167,
      "p95_latency_ms": 4872.095974441618,
      "input_tokens": 492,
      "output_tokens": 21,
      "reasoning_tokens": 134,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0a79b09d-7935-45f3-8451-f3ad084f46ac",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0013739999999999998,
      "cost_per_success_usd": 0.0004579999999999999,
      "octane_score": 5.14410480349345,
      "ci_low": 4.938864628820963,
      "ci_high": 5.543668122270743,
      "ci_metric": "octane_score",
      "median_latency_ms": 1106.5188329666853,
      "p95_latency_ms": 1576.6974333673713,
      "input_tokens": 282,
      "output_tokens": 81,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e3f809f6-5fef-4444-afab-c1824ac39a3b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001274,
      "cost_per_success_usd": 0.00042466666666666667,
      "octane_score": 6.531397174254319,
      "ci_low": 4.40909090909091,
      "ci_high": 8.970649895178196,
      "ci_metric": "octane_score",
      "median_latency_ms": 1351.5101252123713,
      "p95_latency_ms": 2147.3635508678854,
      "input_tokens": 177,
      "output_tokens": 62,
      "reasoning_tokens": 30,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4bd0ef56-6e8e-4d77-a8ec-c294072a1fae",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0008339999999999999,
      "cost_per_success_usd": 0.000278,
      "octane_score": 6.619904076738609,
      "ci_low": 4.820143884892086,
      "ci_high": 8.428057553956835,
      "ci_metric": "octane_score",
      "median_latency_ms": 914.8594588041306,
      "p95_latency_ms": 951.0383336804807,
      "input_tokens": 312,
      "output_tokens": 21,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c511c1aa-c0fa-4cdd-9a03-2d01def1aa69",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00139,
      "cost_per_success_usd": 0.00046333333333333334,
      "octane_score": 4.65179856115108,
      "ci_low": 3.64080459770115,
      "ci_high": 9.371014492753623,
      "ci_metric": "octane_score",
      "median_latency_ms": 2152.558541856706,
      "p95_latency_ms": 2457.1008793078363,
      "input_tokens": 270,
      "output_tokens": 19,
      "reasoning_tokens": 66,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d0433f20-2bc9-4c55-9d37-bebd7a10e7dd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002602,
      "cost_per_success_usd": 0.0008673333333333333,
      "octane_score": 4.523827824750192,
      "ci_low": 4.383546205860255,
      "ci_high": 4.704636290967227,
      "ci_metric": "octane_score",
      "median_latency_ms": 2765.7042080536485,
      "p95_latency_ms": 4428.196621406823,
      "input_tokens": 216,
      "output_tokens": 27,
      "reasoning_tokens": 190,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fae26727-fa28-46ed-8efc-8b532d031b53",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000696,
      "cost_per_success_usd": 0.000232,
      "octane_score": 7.610632183908046,
      "ci_low": 7.51293103448276,
      "ci_high": 7.65948275862069,
      "ci_metric": "octane_score",
      "median_latency_ms": 1836.4143334329128,
      "p95_latency_ms": 2294.496483821422,
      "input_tokens": 273,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bb2b0fe3-d14e-44d4-97cb-3f87a936fe8c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.001092,
      "cost_per_success_usd": 0.000364,
      "octane_score": 5.758241758241758,
      "ci_low": 5.758241758241758,
      "ci_high": 5.758241758241758,
      "ci_metric": "octane_score",
      "median_latency_ms": 1096.3839581236243,
      "p95_latency_ms": 1443.7001834623513,
      "input_tokens": 456,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "62fd8a3e-2ee3-4ae1-a214-fa4331baa72e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.021491999999999997,
      "cost_per_success_usd": 0.007163999999999999,
      "octane_score": 0.29257398101619214,
      "ci_low": 0.18289703315881328,
      "ci_high": 0.41786283891547055,
      "ci_metric": "octane_score",
      "median_latency_ms": 2517.3093751072884,
      "p95_latency_ms": 2831.71721175313,
      "input_tokens": 6282,
      "output_tokens": 99,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c78cc9ea-23b4-43ef-84e0-e85d2d03be52",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.018092999999999998,
      "cost_per_success_usd": 0.006030999999999999,
      "octane_score": 0.2927651577958327,
      "ci_low": 0.17421476730800856,
      "ci_high": 0.4465946217642624,
      "ci_metric": "octane_score",
      "median_latency_ms": 2456.412957981229,
      "p95_latency_ms": 2930.4401461035013,
      "input_tokens": 5541,
      "output_tokens": 90,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3cae4f87-bccf-4942-9e87-1acc64334723",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.016695,
      "cost_per_success_usd": 0.0055650000000000005,
      "octane_score": 0.31686133572926023,
      "ci_low": 0.18454561311704168,
      "ci_high": 0.5796451914098973,
      "ci_metric": "octane_score",
      "median_latency_ms": 2666.152125224471,
      "p95_latency_ms": 2962.0152374729514,
      "input_tokens": 5031,
      "output_tokens": 117,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "208e892b-b1ad-4185-a2db-ef4b370f5ada",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.049890000000000004,
      "cost_per_success_usd": 0.016630000000000002,
      "octane_score": 0.23593906594507918,
      "ci_low": 0.23161969035331478,
      "ci_high": 0.24034412955465587,
      "ci_metric": "octane_score",
      "median_latency_ms": 4149.315124377608,
      "p95_latency_ms": 4369.195062294602,
      "input_tokens": 5856,
      "output_tokens": 126,
      "reasoning_tokens": 561,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4eeb2444-54fa-4c66-abb8-cee161d20fb7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.024564000000000002,
      "cost_per_success_usd": 0.008188,
      "octane_score": 0.2877381533952124,
      "ci_low": 0.195890045369629,
      "ci_high": 0.405554635304511,
      "ci_metric": "octane_score",
      "median_latency_ms": 2530.7625830173492,
      "p95_latency_ms": 3078.423945605755,
      "input_tokens": 5766,
      "output_tokens": 288,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8ccdb3d5-562c-4d04-9710-21d425ffede0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.027459000000000004,
      "cost_per_success_usd": 0.009153000000000001,
      "octane_score": 0.23547834953931318,
      "ci_low": 0.18803799346987238,
      "ci_high": 0.4264725435447413,
      "ci_metric": "octane_score",
      "median_latency_ms": 3787.573833949864,
      "p95_latency_ms": 3947.501846496016,
      "input_tokens": 5559,
      "output_tokens": 96,
      "reasoning_tokens": 98,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e739c26b-4755-4f81-923c-ad19a40e3ac5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0282535,
      "cost_per_success_usd": 0.009417833333333334,
      "octane_score": 0.2945121843311449,
      "ci_low": 0.22461538461538472,
      "ci_high": 0.5714776278287148,
      "ci_metric": "octane_score",
      "median_latency_ms": 3076.077874749899,
      "p95_latency_ms": 4034.0932628139853,
      "input_tokens": 5312,
      "output_tokens": 227,
      "reasoning_tokens": 36,
      "undefined_reason": null
    },
    {
      "snapshot_id": "363d3d0e-b872-4fd0-b3eb-e242092a169c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.019332000000000002,
      "cost_per_success_usd": 0.006444000000000001,
      "octane_score": 0.2855886612869853,
      "ci_low": 0.17352899575671857,
      "ci_high": 0.4983766853825293,
      "ci_metric": "octane_score",
      "median_latency_ms": 3564.1877502202988,
      "p95_latency_ms": 3697.8252621367574,
      "input_tokens": 5715,
      "output_tokens": 108,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ac01c4ff-cbdd-4645-aabc-798363ee2433",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.048964999999999995,
      "cost_per_success_usd": 0.016321666666666665,
      "octane_score": 0.4161135504952517,
      "ci_low": 0.40391084093211754,
      "ci_high": 0.43217338217338225,
      "ci_metric": "octane_score",
      "median_latency_ms": 5675.722415558994,
      "p95_latency_ms": 5852.352254465222,
      "input_tokens": 6781,
      "output_tokens": 108,
      "reasoning_tokens": 394,
      "undefined_reason": null
    },
    {
      "snapshot_id": "23b31f72-62f3-4f37-9cb0-ca615935bcdd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0028079999999999997,
      "cost_per_success_usd": 0.0009359999999999999,
      "octane_score": 4.191951566951568,
      "ci_low": 3.852287581699346,
      "ci_high": 4.515046296296297,
      "ci_metric": "octane_score",
      "median_latency_ms": 2522.067374549806,
      "p95_latency_ms": 2814.2370380461216,
      "input_tokens": 216,
      "output_tokens": 27,
      "reasoning_tokens": 171,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9e393690-01f2-4139-a25b-796cd7d7bfb9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000726,
      "cost_per_success_usd": 0.00024199999999999997,
      "octane_score": 7.296143250688706,
      "ci_low": 7.202479338842977,
      "ci_high": 7.342975206611571,
      "ci_metric": "octane_score",
      "median_latency_ms": 706.5970404073596,
      "p95_latency_ms": 815.6530412845314,
      "input_tokens": 273,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8667fd50-21f9-498d-9f99-93817aff2c6a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0028079999999999997,
      "cost_per_success_usd": 0.0009359999999999999,
      "octane_score": 7.256054131054132,
      "ci_low": 7.008790436005626,
      "ci_high": 7.563034188034189,
      "ci_metric": "octane_score",
      "median_latency_ms": 2266.507999971509,
      "p95_latency_ms": 3088.3257495239372,
      "input_tokens": 492,
      "output_tokens": 21,
      "reasoning_tokens": 131,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d7cf8a1f-a440-496d-b824-3c0955831a57",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00057,
      "cost_per_success_usd": 0.00018999999999999998,
      "octane_score": 9.280701754385966,
      "ci_low": 6.3578947368421055,
      "ci_high": 11.189473684210531,
      "ci_metric": "octane_score",
      "median_latency_ms": 738.6477505788207,
      "p95_latency_ms": 1266.689299046993,
      "input_tokens": 141,
      "output_tokens": 24,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "318278df-2ac2-4782-ab4a-2fcd9792331c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.001536,
      "cost_per_success_usd": 0.000512,
      "octane_score": 4.601562499999999,
      "ci_low": 4.417968750000002,
      "ci_high": 4.958984375,
      "ci_metric": "octane_score",
      "median_latency_ms": 939.667584374547,
      "p95_latency_ms": 991.7400455102324,
      "input_tokens": 282,
      "output_tokens": 81,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "de6288eb-b284-4c87-8720-5f333b1196b5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002196,
      "cost_per_success_usd": 0.000732,
      "octane_score": 2.944444444444445,
      "ci_low": 2.848360655737705,
      "ci_high": 3.0372928176795573,
      "ci_metric": "octane_score",
      "median_latency_ms": 2314.023207873106,
      "p95_latency_ms": 2726.5008195303376,
      "input_tokens": 270,
      "output_tokens": 21,
      "reasoning_tokens": 117,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6c468ec9-0466-4717-93ee-01a7152dac48",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0015600000000000002,
      "cost_per_success_usd": 0.0005200000000000001,
      "octane_score": 4.030769230769231,
      "ci_low": 2.594059405940594,
      "ci_high": 5.574468085106383,
      "ci_metric": "octane_score",
      "median_latency_ms": 2080.697374418378,
      "p95_latency_ms": 2229.975349176675,
      "input_tokens": 456,
      "output_tokens": 20,
      "reasoning_tokens": 34,
      "undefined_reason": null
    },
    {
      "snapshot_id": "05723034-2a44-4a22-a498-bbe401e5982e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0014039999999999999,
      "cost_per_success_usd": 0.00046799999999999994,
      "octane_score": 3.932336182336183,
      "ci_low": 2.8632478632478633,
      "ci_high": 5.006410256410257,
      "ci_metric": "octane_score",
      "median_latency_ms": 975.064542144537,
      "p95_latency_ms": 980.2279906347394,
      "input_tokens": 312,
      "output_tokens": 27,
      "reasoning_tokens": 38,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e4d23596-cba3-4b84-862b-2e071a9385e6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002358,
      "cost_per_success_usd": 0.0007859999999999999,
      "octane_score": 3.5288379983036484,
      "ci_low": 3.376775271512115,
      "ci_high": 3.685615848406546,
      "ci_metric": "octane_score",
      "median_latency_ms": 2299.5234169065952,
      "p95_latency_ms": 2668.742603808641,
      "input_tokens": 177,
      "output_tokens": 66,
      "reasoning_tokens": 101,
      "undefined_reason": null
    },
    {
      "snapshot_id": "af52eb0d-e579-4af8-8cdb-f8adfcde254e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.009435,
      "cost_per_success_usd": 0.0031450000000000002,
      "octane_score": 0.8819289878113409,
      "ci_low": 0.8441971383147855,
      "ci_high": 0.9195548489666137,
      "ci_metric": "octane_score",
      "median_latency_ms": 3778.763917274773,
      "p95_latency_ms": 3840.9262041561306,
      "input_tokens": 5547,
      "output_tokens": 249,
      "reasoning_tokens": 399,
      "undefined_reason": null
    },
    {
      "snapshot_id": "808931a7-294a-446c-a2e3-a2423b777f03",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007594999999999999,
      "cost_per_success_usd": 0.0025316666666666664,
      "octane_score": 0.7269256089532588,
      "ci_low": 0.532739927667867,
      "ci_high": 0.9237744775923248,
      "ci_metric": "octane_score",
      "median_latency_ms": 2202.4020412936807,
      "p95_latency_ms": 2209.7560917958617,
      "input_tokens": 5795,
      "output_tokens": 132,
      "reasoning_tokens": 168,
      "undefined_reason": null
    },
    {
      "snapshot_id": "788e1586-9e01-4431-942a-e2667096a205",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.008889999999999999,
      "cost_per_success_usd": 0.002963333333333333,
      "octane_score": 0.7273340832395953,
      "ci_low": 0.704868154158215,
      "ci_high": 0.7478743906586556,
      "ci_metric": "octane_score",
      "median_latency_ms": 3068.002250045538,
      "p95_latency_ms": 3237.0300489477813,
      "input_tokens": 5830,
      "output_tokens": 114,
      "reasoning_tokens": 396,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1c87c5f9-8f94-41d9-9cb6-a96ff28415f7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0056868000000000005,
      "cost_per_success_usd": 0.0018956,
      "octane_score": 1.242878244355349,
      "ci_low": 0.9095415117719949,
      "ci_high": 1.6482653381805121,
      "ci_metric": "octane_score",
      "median_latency_ms": 2201.6575830057263,
      "p95_latency_ms": 2745.598569978028,
      "input_tokens": 5766,
      "output_tokens": 294,
      "reasoning_tokens": 123,
      "undefined_reason": null
    },
    {
      "snapshot_id": "47cf4542-4d40-4b54-8367-398758b20ffc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.010435,
      "cost_per_success_usd": 0.0034783333333333333,
      "octane_score": 1.1280306660277912,
      "ci_low": 1.0468694414799005,
      "ci_high": 1.1934638034318206,
      "ci_metric": "octane_score",
      "median_latency_ms": 3430.8867501094937,
      "p95_latency_ms": 4004.2673748917878,
      "input_tokens": 5809,
      "output_tokens": 132,
      "reasoning_tokens": 639,
      "undefined_reason": null
    },
    {
      "snapshot_id": "10457d7e-052d-43f5-bd85-a3c50fa6bf42",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003603,
      "cost_per_success_usd": 0.001201,
      "octane_score": 1.4682209270052733,
      "ci_low": 0.8759728431859579,
      "ci_high": 2.6944444444444446,
      "ci_metric": "octane_score",
      "median_latency_ms": 1812.6926245167851,
      "p95_latency_ms": 2249.8448634520173,
      "input_tokens": 5031,
      "output_tokens": 121,
      "reasoning_tokens": 40,
      "undefined_reason": null
    },
    {
      "snapshot_id": "be24529f-1278-40bd-8101-ad939b16a8ad",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007337,
      "cost_per_success_usd": 0.0024456666666666667,
      "octane_score": 0.8570260324383263,
      "ci_low": 0.8397435897435899,
      "ci_high": 0.8711554447215295,
      "ci_metric": "octane_score",
      "median_latency_ms": 1825.063000433147,
      "p95_latency_ms": 2023.0080249719322,
      "input_tokens": 6293,
      "output_tokens": 109,
      "reasoning_tokens": 65,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a16680c6-7556-4524-885d-f9d7d469186b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.010423,
      "cost_per_success_usd": 0.003474333333333333,
      "octane_score": 1.9548114746234289,
      "ci_low": 1.9047300525561397,
      "ci_high": 2.031408384693235,
      "ci_metric": "octane_score",
      "median_latency_ms": 3677.7514591813087,
      "p95_latency_ms": 4852.995421551168,
      "input_tokens": 6775,
      "output_tokens": 114,
      "reasoning_tokens": 494,
      "undefined_reason": null
    },
    {
      "snapshot_id": "76a16375-6c3b-4f59-be4a-1d0f71a2e4bc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0040566,
      "cost_per_success_usd": 0.0013521999999999998,
      "octane_score": 1.3057733076961004,
      "ci_low": 0.7914238756910205,
      "ci_high": 1.9798908574418776,
      "ci_metric": "octane_score",
      "median_latency_ms": 1888.9424167573452,
      "p95_latency_ms": 1916.894579399377,
      "input_tokens": 5541,
      "output_tokens": 96,
      "reasoning_tokens": 67,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a386fae3-1e28-442d-9eb1-e5160ba56c44",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 5.521e-05,
      "cost_per_success_usd": 1.8403333333333334e-05,
      "octane_score": 100.0,
      "ci_low": 1.3399999999999999e-05,
      "ci_high": 2.3429999999999998e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 1434.5296677201986,
      "p95_latency_ms": 2531.4530036412175,
      "input_tokens": 469,
      "output_tokens": 19,
      "reasoning_tokens": 223,
      "undefined_reason": null
    },
    {
      "snapshot_id": "703c6ca3-d899-4ca3-8ad0-dd29df2500cd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 8.321000000000001e-05,
      "cost_per_success_usd": 2.773666666666667e-05,
      "octane_score": 100.0,
      "ci_low": 2.6550000000000005e-05,
      "ci_high": 2.892e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 4898.147500120103,
      "p95_latency_ms": 5369.187999889255,
      "input_tokens": 371,
      "output_tokens": 212,
      "reasoning_tokens": 212,
      "undefined_reason": null
    },
    {
      "snapshot_id": "87673ffa-41c7-4bdb-b72b-115a140e4674",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 5.297e-05,
      "cost_per_success_usd": 1.7656666666666666e-05,
      "octane_score": 100.0,
      "ci_low": 1.743e-05,
      "ci_high": 1.777e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 813.5855831205845,
      "p95_latency_ms": 1955.7864202186463,
      "input_tokens": 451,
      "output_tokens": 39,
      "reasoning_tokens": 193,
      "undefined_reason": null
    },
    {
      "snapshot_id": "914ca41f-6d2f-4cdc-9672-ace9fa4a5ac6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00011771000000000001,
      "cost_per_success_usd": 3.923666666666667e-05,
      "octane_score": 100.00000000000001,
      "ci_low": 3.884e-05,
      "ci_high": 3.986e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 2450.4117500036955,
      "p95_latency_ms": 3830.5209887214005,
      "input_tokens": 399,
      "output_tokens": 203,
      "reasoning_tokens": 419,
      "undefined_reason": null
    },
    {
      "snapshot_id": "86f2cb0b-68ff-4d7d-ad5f-b3249c5ca65c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00020375,
      "cost_per_success_usd": 6.791666666666666e-05,
      "octane_score": 100.0,
      "ci_low": 6.637e-05,
      "ci_high": 7.079e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 10264.75316658616,
      "p95_latency_ms": 10898.982903920114,
      "input_tokens": 649,
      "output_tokens": 447,
      "reasoning_tokens": 637,
      "undefined_reason": null
    },
    {
      "snapshot_id": "60830e87-c6d6-4249-a9d7-92390696888d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 7.068e-05,
      "cost_per_success_usd": 2.3559999999999998e-05,
      "octane_score": 99.99999999999999,
      "ci_low": 2.2620000000000007e-05,
      "ci_high": 2.539e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 3769.7512498125434,
      "p95_latency_ms": 4231.195612065494,
      "input_tokens": 452,
      "output_tokens": 122,
      "reasoning_tokens": 214,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5a13bcc6-3c5c-4f9f-bfaf-e63c480e3b6b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 5.29e-05,
      "cost_per_success_usd": 1.7633333333333333e-05,
      "octane_score": 100.0,
      "ci_low": 1.208e-05,
      "ci_high": 2.1260000000000007e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 2571.419416926801,
      "p95_latency_ms": 3816.777992621064,
      "input_tokens": 324,
      "output_tokens": 30,
      "reasoning_tokens": 224,
      "undefined_reason": null
    },
    {
      "snapshot_id": "133472bf-010d-4789-8124-ac74ad9fffec",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 6.466000000000001e-05,
      "cost_per_success_usd": 2.1553333333333336e-05,
      "octane_score": 100.00000000000001,
      "ci_low": 2.0850000000000002e-05,
      "ci_high": 2.216e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 3325.58283302933,
      "p95_latency_ms": 3685.936869867146,
      "input_tokens": 427,
      "output_tokens": 123,
      "reasoning_tokens": 182,
      "undefined_reason": null
    },
    {
      "snapshot_id": "66259c02-8f41-4b94-be2e-9f1b1fa03131",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 6.288000000000001e-05,
      "cost_per_success_usd": 2.0960000000000003e-05,
      "octane_score": 100.0,
      "ci_low": 2.0960000000000003e-05,
      "ci_high": 2.0960000000000003e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 2236.01745814085,
      "p95_latency_ms": 2460.208544973284,
      "input_tokens": 600,
      "output_tokens": 57,
      "reasoning_tokens": 207,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d1fb72e6-1eaa-4077-a84e-201cdc764428",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0045056,
      "cost_per_success_usd": 0.0015018666666666667,
      "octane_score": 1.4351029829545456,
      "ci_low": 1.3171188788397294,
      "ci_high": 1.5314792459838427,
      "ci_metric": "octane_score",
      "median_latency_ms": 3957.4646251276135,
      "p95_latency_ms": 4450.211137719452,
      "input_tokens": 873,
      "output_tokens": 3,
      "reasoning_tokens": 602,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2c4b29f0-32b2-4db0-8481-8112edea9fe5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0047216,
      "cost_per_success_usd": 0.0015738666666666667,
      "octane_score": 1.1693070145713318,
      "ci_low": 0.8623648532692637,
      "ci_high": 1.488690274483226,
      "ci_metric": "octane_score",
      "median_latency_ms": 4010.242166928947,
      "p95_latency_ms": 4231.736928783357,
      "input_tokens": 918,
      "output_tokens": 6,
      "reasoning_tokens": 620,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1e699b53-6cfe-44c3-bdb9-8ff51854628d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004036,
      "cost_per_success_usd": 0.0013453333333333331,
      "octane_score": 2.06169474727453,
      "ci_low": 1.8568164863856778,
      "ci_high": 2.2625846023688663,
      "ci_metric": "octane_score",
      "median_latency_ms": 3218.4427501633763,
      "p95_latency_ms": 3673.7194499000907,
      "input_tokens": 786,
      "output_tokens": 60,
      "reasoning_tokens": 532,
      "undefined_reason": null
    },
    {
      "snapshot_id": "224b71f3-1db1-4bad-a852-2fc77abb954b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0034,
      "cost_per_success_usd": 0.0011333333333333332,
      "octane_score": 1.5579411764705884,
      "ci_low": 1.4548319327731092,
      "ci_high": 1.7856661272923409,
      "ci_metric": "octane_score",
      "median_latency_ms": 3162.892749533057,
      "p95_latency_ms": 3331.185475643724,
      "input_tokens": 885,
      "output_tokens": 3,
      "reasoning_tokens": 450,
      "undefined_reason": null
    },
    {
      "snapshot_id": "695fe4b5-9a00-47a1-b6ab-60c4aaec5d11",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0049452,
      "cost_per_success_usd": 0.0016484000000000002,
      "octane_score": 1.271536034942975,
      "ci_low": 1.2032146957520093,
      "ci_high": 1.4040728831725615,
      "ci_metric": "octane_score",
      "median_latency_ms": 3097.4243758246303,
      "p95_latency_ms": 3258.5229127667844,
      "input_tokens": 1038,
      "output_tokens": 6,
      "reasoning_tokens": 581,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6e1f08dd-1ee9-4fa6-b562-ad68faa82290",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0046435999999999995,
      "cost_per_success_usd": 0.0015478666666666665,
      "octane_score": 1.5220949263502455,
      "ci_low": 1.3577879005833933,
      "ci_high": 1.7519333729922664,
      "ci_metric": "octane_score",
      "median_latency_ms": 3172.831540927291,
      "p95_latency_ms": 3393.103204574436,
      "input_tokens": 879,
      "output_tokens": 108,
      "reasoning_tokens": 518,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4212ef70-5740-43be-9b55-2d20dc66d873",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.007297199999999999,
      "cost_per_success_usd": 0.0024324,
      "octane_score": 1.6130844707559067,
      "ci_low": 1.5886565376027881,
      "ci_high": 1.6387107383654005,
      "ci_metric": "octane_score",
      "median_latency_ms": 5969.854750670493,
      "p95_latency_ms": 6291.35762527585,
      "input_tokens": 825,
      "output_tokens": 9,
      "reasoning_tokens": 1041,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7d3c117e-1a49-413e-9c81-37b63fb5c313",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00811,
      "cost_per_success_usd": 0.002703333333333333,
      "octane_score": 2.5123304562268802,
      "ci_low": 2.0777247511829007,
      "ci_high": 2.911379743941472,
      "ci_metric": "octane_score",
      "median_latency_ms": 5014.317249879241,
      "p95_latency_ms": 7443.239537626505,
      "input_tokens": 1110,
      "output_tokens": 3,
      "reasoning_tokens": 1160,
      "undefined_reason": null
    },
    {
      "snapshot_id": "20ec6bbc-58a9-4a01-8689-c70d6ddb0a38",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0024552,
      "cost_per_success_usd": 0.0008183999999999999,
      "octane_score": 2.154610622352558,
      "ci_low": 1.4760508308895406,
      "ci_high": 2.5977517106549377,
      "ci_metric": "octane_score",
      "median_latency_ms": 1998.2072096318007,
      "p95_latency_ms": 2350.828970130533,
      "input_tokens": 744,
      "output_tokens": 12,
      "reasoning_tokens": 258,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d2877cc6-4cf7-48cf-b12e-3eec14495103",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.005424,
      "cost_per_success_usd": 0.001808,
      "octane_score": 1.1592920353982303,
      "ci_low": 1.0543259557344065,
      "ci_high": 1.2922318125770655,
      "ci_metric": "octane_score",
      "median_latency_ms": 3555.5621664971113,
      "p95_latency_ms": 3656.0032171197236,
      "input_tokens": 1038,
      "output_tokens": 6,
      "reasoning_tokens": 680,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2bdba4d4-e70a-4214-9c5e-98510e47d52a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009984,
      "cost_per_success_usd": 0.003328,
      "octane_score": 2.0407652243589745,
      "ci_low": 1.9565174715351397,
      "ci_high": 2.128527607361963,
      "ci_metric": "octane_score",
      "median_latency_ms": 7136.185749433935,
      "p95_latency_ms": 7877.348875440657,
      "input_tokens": 1110,
      "output_tokens": 3,
      "reasoning_tokens": 1419,
      "undefined_reason": null
    },
    {
      "snapshot_id": "49615fd1-0949-4c6b-83af-df1250487f46",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004254,
      "cost_per_success_usd": 0.001418,
      "octane_score": 1.2451810061118946,
      "ci_low": 0.9897234678624813,
      "ci_high": 1.4400324149108592,
      "ci_metric": "octane_score",
      "median_latency_ms": 2756.894290447235,
      "p95_latency_ms": 3677.683766465634,
      "input_tokens": 885,
      "output_tokens": 3,
      "reasoning_tokens": 571,
      "undefined_reason": null
    },
    {
      "snapshot_id": "43bc6ba5-d4bf-4c8d-9b06-23eaf1b75c77",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0045,
      "cost_per_success_usd": 0.0014999999999999998,
      "octane_score": 1.8491111111111118,
      "ci_low": 1.685297691373026,
      "ci_high": 2.0334799608993164,
      "ci_metric": "octane_score",
      "median_latency_ms": 2994.1806253045797,
      "p95_latency_ms": 3391.073537245393,
      "input_tokens": 786,
      "output_tokens": 60,
      "reasoning_tokens": 556,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ee78afd5-1f84-43f0-b8b9-cc19f8e3de4b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00489,
      "cost_per_success_usd": 0.0016300000000000002,
      "octane_score": 1.129038854805726,
      "ci_low": 0.8241082410824107,
      "ci_high": 1.43390452876377,
      "ci_metric": "octane_score",
      "median_latency_ms": 2670.625957660377,
      "p95_latency_ms": 3018.777983635664,
      "input_tokens": 918,
      "output_tokens": 6,
      "reasoning_tokens": 663,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a898b18d-2c23-4977-b719-0ad71ff3fe48",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.005039999999999999,
      "cost_per_success_usd": 0.0016799999999999999,
      "octane_score": 1.2829365079365083,
      "ci_low": 1.2024221453287196,
      "ci_high": 1.3624535315985131,
      "ci_metric": "octane_score",
      "median_latency_ms": 3408.845209516585,
      "p95_latency_ms": 3524.0696968510747,
      "input_tokens": 873,
      "output_tokens": 3,
      "reasoning_tokens": 706,
      "undefined_reason": null
    },
    {
      "snapshot_id": "464037fa-b7e0-406e-8e80-43c9e3980530",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.006096000000000001,
      "cost_per_success_usd": 0.0020320000000000004,
      "octane_score": 1.1594488188976373,
      "ci_low": 1.0954223081882652,
      "ci_high": 1.2453496985555224,
      "ci_metric": "octane_score",
      "median_latency_ms": 3472.5848753005266,
      "p95_latency_ms": 3778.4337871707976,
      "input_tokens": 879,
      "output_tokens": 93,
      "reasoning_tokens": 790,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7751d68d-39e3-41e6-bdd5-9199b11d2628",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.008694,
      "cost_per_success_usd": 0.002898,
      "octane_score": 1.3539222452265933,
      "ci_low": 1.227149698105351,
      "ci_high": 1.5732424485431704,
      "ci_metric": "octane_score",
      "median_latency_ms": 4778.2547092065215,
      "p95_latency_ms": 5058.441996201873,
      "input_tokens": 825,
      "output_tokens": 9,
      "reasoning_tokens": 1293,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8994a7f6-d2c2-4921-afe3-5c703b4cba43",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003228,
      "cost_per_success_usd": 0.001076,
      "octane_score": 1.6387856257744735,
      "ci_low": 1.1615384615384614,
      "ci_high": 2.0944669365722,
      "ci_metric": "octane_score",
      "median_latency_ms": 2663.733291439712,
      "p95_latency_ms": 3184.288341924548,
      "input_tokens": 744,
      "output_tokens": 12,
      "reasoning_tokens": 374,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ab6becec-8bc1-4a93-9b19-5122a268bd05",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00477072,
      "cost_per_success_usd": 0.00159024,
      "octane_score": 1.1572676660965222,
      "ci_low": 0.8345524670833886,
      "ci_high": 1.580518247184914,
      "ci_metric": "octane_score",
      "median_latency_ms": 11025.402208790183,
      "p95_latency_ms": 12864.837019704282,
      "input_tokens": 324,
      "output_tokens": 15,
      "reasoning_tokens": 2077,
      "undefined_reason": null
    },
    {
      "snapshot_id": "eb4bc65d-2dc3-4f16-b6dd-a3caf8ca1686",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0027036,
      "cost_per_success_usd": 0.0009012,
      "octane_score": 2.614292055037727,
      "ci_low": 2.23010943507838,
      "ci_high": 3.245849693462836,
      "ci_metric": "octane_score",
      "median_latency_ms": 7654.391082935035,
      "p95_latency_ms": 9049.428321048617,
      "input_tokens": 293,
      "output_tokens": 97,
      "reasoning_tokens": 1046,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2fd91a47-8415-423c-86f0-fcc3c0d96e7b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0104049,
      "cost_per_success_usd": 0.0034683,
      "octane_score": 1.1312939096002845,
      "ci_low": 0.46399127753078545,
      "ci_high": 4.086655291867018,
      "ci_metric": "octane_score",
      "median_latency_ms": 7812.817084603012,
      "p95_latency_ms": 46569.46312114596,
      "input_tokens": 237,
      "output_tokens": 987,
      "reasoning_tokens": 3590,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7320aa3b-784a-43cf-840c-66b69923415f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0009495,
      "cost_per_success_usd": 0.0003165,
      "octane_score": 5.5713533438651925,
      "ci_low": 3.8167456556082144,
      "ci_high": 6.664519592055824,
      "ci_metric": "octane_score",
      "median_latency_ms": 2833.9490005746484,
      "p95_latency_ms": 3105.7906629517674,
      "input_tokens": 150,
      "output_tokens": 16,
      "reasoning_tokens": 376,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ea15df44-7fa5-4780-8490-dcf808c85f0a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.00213075,
      "cost_per_success_usd": 0.0007102499999999999,
      "octane_score": 3.905197700340257,
      "ci_low": 3.7201168864039036,
      "ci_high": 4.097775862481745,
      "ci_metric": "octane_score",
      "median_latency_ms": 5358.849042095244,
      "p95_latency_ms": 5823.343391995877,
      "input_tokens": 185,
      "output_tokens": 62,
      "reasoning_tokens": 848,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ec08b444-0d88-4aee-9b70-1edf0b86a5aa",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00292888,
      "cost_per_success_usd": 0.0009762933333333333,
      "octane_score": 2.207669826008577,
      "ci_low": 1.9197346483088542,
      "ci_high": 2.6051819363222872,
      "ci_metric": "octane_score",
      "median_latency_ms": 6513.6895421892405,
      "p95_latency_ms": 6577.356891147792,
      "input_tokens": 283,
      "output_tokens": 111,
      "reasoning_tokens": 1150,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f9cbfc53-01ca-4edc-b084-eca0310bccae",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00105269,
      "cost_per_success_usd": 0.00035089666666666665,
      "octane_score": 5.03187073117442,
      "ci_low": 4.187517293173642,
      "ci_high": 5.692210150767809,
      "ci_metric": "octane_score",
      "median_latency_ms": 3135.9373331069946,
      "p95_latency_ms": 3268.851921800524,
      "input_tokens": 277,
      "output_tokens": 7,
      "reasoning_tokens": 421,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a5ffebd3-324e-4006-9b5a-01b43ec161eb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0040383,
      "cost_per_success_usd": 0.0013461,
      "octane_score": 5.045439912834608,
      "ci_low": 2.8636742338173833,
      "ci_high": 8.606511627906976,
      "ci_metric": "octane_score",
      "median_latency_ms": 6409.872000105679,
      "p95_latency_ms": 14474.430524837224,
      "input_tokens": 509,
      "output_tokens": 5,
      "reasoning_tokens": 1688,
      "undefined_reason": null
    },
    {
      "snapshot_id": "08088c5c-2c92-46d3-b85a-8c7dea407508",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00327188,
      "cost_per_success_usd": 0.0010906266666666667,
      "octane_score": 1.921830873992934,
      "ci_low": 1.6593437042314847,
      "ci_high": 2.3471707409937403,
      "ci_metric": "octane_score",
      "median_latency_ms": 8486.203375272453,
      "p95_latency_ms": 9245.812675449997,
      "input_tokens": 471,
      "output_tokens": 12,
      "reasoning_tokens": 1401,
      "undefined_reason": null
    },
    {
      "snapshot_id": "89e0c238-cf57-4ddd-92e6-30511f3006bb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0060804,
      "cost_per_success_usd": 0.0020268,
      "octane_score": 1.034142490625617,
      "ci_low": 0.9285840864788234,
      "ci_high": 1.0964636953337519,
      "ci_metric": "octane_score",
      "median_latency_ms": 1137.6242907717824,
      "p95_latency_ms": 1249.2503541521728,
      "input_tokens": 705,
      "output_tokens": 54,
      "reasoning_tokens": 291,
      "undefined_reason": null
    },
    {
      "snapshot_id": "49c3f509-bbc3-402a-92c8-ed074345a77e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004905,
      "cost_per_success_usd": 0.001635,
      "octane_score": 2.3997961264016316,
      "ci_low": 2.2450767531329756,
      "ci_high": 2.5089810236253745,
      "ci_metric": "octane_score",
      "median_latency_ms": 2573.943833820522,
      "p95_latency_ms": 3281.4820961095393,
      "input_tokens": 468,
      "output_tokens": 135,
      "reasoning_tokens": 156,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9aaeec13-32e9-4e06-bc97-b4ab1f2f2b6b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0053952,
      "cost_per_success_usd": 0.0017984000000000001,
      "octane_score": 0.9805011862396203,
      "ci_low": 0.6801801801801801,
      "ci_high": 1.1851851851851853,
      "ci_metric": "octane_score",
      "median_latency_ms": 1174.5395427569747,
      "p95_latency_ms": 1304.827329330146,
      "input_tokens": 221,
      "output_tokens": 231,
      "reasoning_tokens": 96,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8493c22d-a161-4c91-9325-c2d4245a3e99",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.010101,
      "cost_per_success_usd": 0.003367,
      "octane_score": 0.6997326997326996,
      "ci_low": 0.6610017714964279,
      "ci_high": 0.7485259433962264,
      "ci_metric": "octane_score",
      "median_latency_ms": 3167.578624561429,
      "p95_latency_ms": 3292.525549605489,
      "input_tokens": 525,
      "output_tokens": 142,
      "reasoning_tokens": 484,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f3f2fa93-20ee-4dbc-bbb2-6e62b551a171",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004284,
      "cost_per_success_usd": 0.0014279999999999998,
      "octane_score": 1.509337068160598,
      "ci_low": 1.4034733441033924,
      "ci_high": 1.6046409807355517,
      "ci_metric": "octane_score",
      "median_latency_ms": 754.6760002151132,
      "p95_latency_ms": 1164.3996118567884,
      "input_tokens": 516,
      "output_tokens": 117,
      "reasoning_tokens": 123,
      "undefined_reason": null
    },
    {
      "snapshot_id": "edadef20-14cb-40aa-ac3e-584d6163bf08",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00465,
      "cost_per_success_usd": 0.00155,
      "octane_score": 1.1873118279569892,
      "ci_low": 0.8645161290322579,
      "ci_high": 1.5843411530428873,
      "ci_metric": "octane_score",
      "median_latency_ms": 1905.9110414236784,
      "p95_latency_ms": 2201.6828789375722,
      "input_tokens": 558,
      "output_tokens": 36,
      "reasoning_tokens": 220,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d2561adf-bf64-4f6c-94d6-eaf9222b782b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.005427,
      "cost_per_success_usd": 0.001809,
      "octane_score": 1.5332596277869912,
      "ci_low": 1.44146485119357,
      "ci_high": 1.6287922043317724,
      "ci_metric": "octane_score",
      "median_latency_ms": 1096.8918753787875,
      "p95_latency_ms": 1154.5698013156652,
      "input_tokens": 417,
      "output_tokens": 177,
      "reasoning_tokens": 159,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b8199665-18a6-4993-8212-e1f1ed6c5eb9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003816,
      "cost_per_success_usd": 0.001272,
      "octane_score": 1.3881027253668765,
      "ci_low": 1.2728277585544023,
      "ci_high": 1.463274044795784,
      "ci_metric": "octane_score",
      "median_latency_ms": 827.3042496293783,
      "p95_latency_ms": 937.4818379990757,
      "input_tokens": 510,
      "output_tokens": 81,
      "reasoning_tokens": 129,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5eabd9b6-768e-46a7-9520-5aee779c7004",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0111834,
      "cost_per_success_usd": 0.0037278,
      "octane_score": 1.821896739810791,
      "ci_low": 1.7203214100570245,
      "ci_high": 1.9287858572381589,
      "ci_metric": "octane_score",
      "median_latency_ms": 2670.3514168038964,
      "p95_latency_ms": 2743.93170485273,
      "input_tokens": 741,
      "output_tokens": 322,
      "reasoning_tokens": 356,
      "undefined_reason": null
    },
    {
      "snapshot_id": "69a9c18f-f06a-4dbd-8dad-f1eebe2c6479",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 4.26e-05,
      "cost_per_success_usd": 1.42e-05,
      "octane_score": 124.17840375586854,
      "ci_low": 85.07042253521126,
      "ci_high": 149.71830985915497,
      "ci_metric": "octane_score",
      "median_latency_ms": 199.96870867908,
      "p95_latency_ms": 825.3522331826388,
      "input_tokens": 153,
      "output_tokens": 15,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cba54522-f51e-4def-8b68-80a737eeba1c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.0009022,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3901.8115401268005,
      "p95_latency_ms": 3971.7438663356006,
      "input_tokens": 495,
      "output_tokens": 1004,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "6c6a3d5f-f9b4-4f87-a070-34890fce160e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.0009289999999999999,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 4214.9417493492365,
      "p95_latency_ms": 4400.8304874412715,
      "input_tokens": 225,
      "output_tokens": 1105,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "136564bd-6eaa-4a52-ba4d-de1f0e99ac2d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 1,
      "content_successes": 3,
      "success_rate": 0.3333333333333333,
      "total_cost_usd": 0.000291,
      "cost_per_success_usd": 0.000291,
      "octane_score": 7.202749140893471,
      "ci_low": 7.202749140893471,
      "ci_high": 20.918163672654693,
      "ci_metric": "octane_score",
      "median_latency_ms": 1174.525499343872,
      "p95_latency_ms": 1468.793550506234,
      "input_tokens": 471,
      "output_tokens": 246,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a8eeac12-3655-468f-90f8-59cd6fb0b838",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 6.12e-05,
      "cost_per_success_usd": 2.0399999999999998e-05,
      "octane_score": 105.65359477124186,
      "ci_low": 102.20588235294119,
      "ci_high": 108.62745098039217,
      "ci_metric": "octane_score",
      "median_latency_ms": 454.4007508084178,
      "p95_latency_ms": 630.8706495910882,
      "input_tokens": 282,
      "output_tokens": 6,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "39e61c6b-86b7-4efa-8fdc-487a76c46374",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 8.58e-05,
      "cost_per_success_usd": 2.8599999999999997e-05,
      "octane_score": 96.98135198135202,
      "ci_low": 92.83216783216787,
      "ci_high": 101.11888111888113,
      "ci_metric": "octane_score",
      "median_latency_ms": 359.38045755028725,
      "p95_latency_ms": 404.4620208442211,
      "input_tokens": 189,
      "output_tokens": 60,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "07abbe1d-bc11-490e-92f6-fa7e6cdb16a6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.00018600000000000002,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 706.7728750407696,
      "p95_latency_ms": 968.2105378247797,
      "input_tokens": 282,
      "output_tokens": 162,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "e8670da8-897d-4160-81a7-2010596a64d9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0001716,
      "cost_per_success_usd": 5.72e-05,
      "octane_score": 41.18881118881118,
      "ci_low": 39.545454545454554,
      "ci_high": 44.38811188811189,
      "ci_metric": "octane_score",
      "median_latency_ms": 704.1610423475504,
      "p95_latency_ms": 723.960854485631,
      "input_tokens": 294,
      "output_tokens": 141,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fd063ef5-cda8-4137-93e4-80bac7a0ab00",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.0006864000000000001,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 2982.053459621966,
      "p95_latency_ms": 4960.514083039015,
      "input_tokens": 324,
      "output_tokens": 777,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "eef814f2-04cd-4dd4-80dd-ede4f0ac3bb2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 27.893291786313057,
      "p95_latency_ms": 28.540579229593277,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "3a2ab83c-58c0-4cc4-a2e1-85d9ab7d679f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 29.192499816417694,
      "p95_latency_ms": 33.05233856663108,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "e8b590a2-8eab-49e6-ac07-eaa2a5365290",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.043333768844604,
      "p95_latency_ms": 31.687671225517988,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "002a408d-a1ab-426f-93de-856a86ce2b33",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 31.459415331482887,
      "p95_latency_ms": 41.135729104280465,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "ffddf011-67c3-4262-b361-ff296127ae22",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 28.435499407351017,
      "p95_latency_ms": 30.949874222278595,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "3ecc1d46-f5fe-4048-a318-b8c63491b221",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 29.013833962380886,
      "p95_latency_ms": 32.018670812249184,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "755d8d20-703f-4a70-9a09-ea8f853e32cd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.795332975685596,
      "p95_latency_ms": 33.254584204405546,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "9cfee8fe-735c-4406-a93c-69d190f9e3be",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 31.11241664737463,
      "p95_latency_ms": 33.72672889381647,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "e6a76b42-234a-4321-b7b8-7d51d40378e4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.243958346545696,
      "p95_latency_ms": 30.259520187973976,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "de3063f1-ec75-43b4-89ec-70bb9e7d71c8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.012899999999999998,
      "cost_per_success_usd": 0.004299999999999999,
      "octane_score": 0.5012403100775196,
      "ci_low": 0.4789629629629631,
      "ci_high": 0.5235714285714286,
      "ci_metric": "octane_score",
      "median_latency_ms": 2913.9953339472413,
      "p95_latency_ms": 3061.409070342779,
      "input_tokens": 420,
      "output_tokens": 9,
      "reasoning_tokens": 165,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0213193c-beb0-48a2-a0b7-f9a96b800b37",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.02842,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3496.8114579096437,
      "p95_latency_ms": 4746.365720964968,
      "input_tokens": 687,
      "output_tokens": 313,
      "reasoning_tokens": 118,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "6785f64d-5f99-4f2f-a6de-27430db0d8a0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.014219999999999998,
      "cost_per_success_usd": 0.004739999999999999,
      "octane_score": 0.5851617440225038,
      "ci_low": 0.5601265822784812,
      "ci_high": 0.610126582278481,
      "ci_metric": "octane_score",
      "median_latency_ms": 2837.789417244494,
      "p95_latency_ms": 3213.662940915674,
      "input_tokens": 282,
      "output_tokens": 81,
      "reasoning_tokens": 147,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8043f3e1-0556-45cb-8c1a-7986d70a8957",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.0186,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 2572.397374548018,
      "p95_latency_ms": 2700.25347545743,
      "input_tokens": 405,
      "output_tokens": 205,
      "reasoning_tokens": 86,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "6ab8d2f1-d706-405c-92a2-6d7940a9828a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.01756,
      "cost_per_success_usd": 0.005853333333333333,
      "octane_score": 1.1603075170842825,
      "ci_low": 1.1287089467723668,
      "ci_high": 1.2093963553530753,
      "ci_metric": "octane_score",
      "median_latency_ms": 2255.947041325271,
      "p95_latency_ms": 2573.842342291027,
      "input_tokens": 756,
      "output_tokens": 9,
      "reasoning_tokens": 191,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7f9410bd-1281-4562-a1b5-0407c49b9f7d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00303,
      "cost_per_success_usd": 0.00101,
      "octane_score": 1.7458745874587458,
      "ci_low": 1.1960396039603958,
      "ci_high": 2.1049504950495055,
      "ci_metric": "octane_score",
      "median_latency_ms": 1109.0069999918342,
      "p95_latency_ms": 1157.4085119180381,
      "input_tokens": 213,
      "output_tokens": 18,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bfab877f-97a5-468b-9bd2-9bcac3ee911f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.020880000000000003,
      "cost_per_success_usd": 0.006960000000000001,
      "octane_score": 0.5637452107279693,
      "ci_low": 0.5394822006472492,
      "ci_high": 0.6038699690402476,
      "ci_metric": "octane_score",
      "median_latency_ms": 3004.206291399896,
      "p95_latency_ms": 3078.7353673018515,
      "input_tokens": 378,
      "output_tokens": 21,
      "reasoning_tokens": 321,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d426f8ab-a1c4-4cc2-b280-47eb2e909b80",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.017839999999999998,
      "cost_per_success_usd": 0.005946666666666666,
      "octane_score": 0.39618834080717485,
      "ci_low": 0.36954866775421424,
      "ci_high": 0.4269618834080718,
      "ci_metric": "octane_score",
      "median_latency_ms": 2534.3097914010286,
      "p95_latency_ms": 2907.2779286652803,
      "input_tokens": 474,
      "output_tokens": 174,
      "reasoning_tokens": 88,
      "undefined_reason": null
    },
    {
      "snapshot_id": "41e7a4d6-239b-4595-8c6e-8211db7e89a4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": 3,
      "success_rate": 0.0,
      "total_cost_usd": 0.02901,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3339.725833386183,
      "p95_latency_ms": 3796.4137706905603,
      "input_tokens": 486,
      "output_tokens": 352,
      "reasoning_tokens": 131,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "97fbdbd0-fc07-4fd5-a6d5-b6ca4ad7276c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00080938,
      "cost_per_success_usd": 0.0002697933333333333,
      "octane_score": 25.17358966122217,
      "ci_low": 23.116632629906967,
      "ci_high": 26.985299836664847,
      "ci_metric": "octane_score",
      "median_latency_ms": 4394.101042300463,
      "p95_latency_ms": 5251.427129562943,
      "input_tokens": 564,
      "output_tokens": 2972,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "aa59b012-9f19-4a31-ba12-09443b147717",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00037830000000000003,
      "cost_per_success_usd": 0.0001261,
      "octane_score": 14.002114723764208,
      "ci_low": 13.076426157821505,
      "ci_high": 16.05443413954052,
      "ci_metric": "octane_score",
      "median_latency_ms": 2677.6790004223585,
      "p95_latency_ms": 2847.9073242284358,
      "input_tokens": 312,
      "output_tokens": 1377,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f63242d2-b37b-45fb-a06d-653a513ede8e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.000550095,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 3529.9987075850368,
      "p95_latency_ms": 3974.2295207455754,
      "input_tokens": 207,
      "output_tokens": 2064,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "aa90824d-5d39-4dc2-8cb0-a9bd4f7d12a5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00042724500000000003,
      "cost_per_success_usd": 0.000142415,
      "octane_score": 12.922327938302379,
      "ci_low": 9.40912123020749,
      "ci_high": 16.451918688340413,
      "ci_metric": "octane_score",
      "median_latency_ms": 3475.087375380099,
      "p95_latency_ms": 3689.035037253052,
      "input_tokens": 345,
      "output_tokens": 1557,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b56ef4e8-7f9d-4e2e-a8e5-4b6c909ff608",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0005176600000000001,
      "cost_per_success_usd": 0.00017255333333333336,
      "octane_score": 12.1469690530464,
      "ci_low": 10.864607091022187,
      "ci_high": 13.571613571613572,
      "ci_metric": "octane_score",
      "median_latency_ms": 4050.176165997982,
      "p95_latency_ms": 4213.3402788080275,
      "input_tokens": 504,
      "output_tokens": 1865,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f2fe7ff9-9407-44f7-aaa1-db958861bcc1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000695565,
      "cost_per_success_usd": 0.000231855,
      "octane_score": 9.296039910001223,
      "ci_low": 8.992689396389986,
      "ci_high": 9.557697698992905,
      "ci_metric": "octane_score",
      "median_latency_ms": 3905.6721655651927,
      "p95_latency_ms": 4058.6772298440337,
      "input_tokens": 309,
      "output_tokens": 2598,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9ac85f1d-7032-418b-baf7-4f35ad1e4b96",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 0,
      "content_successes": null,
      "success_rate": 0.0,
      "total_cost_usd": 0.0006981,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 5490.600000135601,
      "p95_latency_ms": 5690.911875478923,
      "input_tokens": 312,
      "output_tokens": 2607,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "f46e4c45-3f3b-4a99-bd7b-d265b0450994",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004596865,
      "cost_per_success_usd": 0.0015322883333333335,
      "octane_score": 2.560658187699661,
      "ci_low": 1.2013609581128808,
      "ci_high": 8.894077289539204,
      "ci_metric": "octane_score",
      "median_latency_ms": 14147.07316737622,
      "p95_latency_ms": 32191.123215947297,
      "input_tokens": 249,
      "output_tokens": 17618,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fa7e6a31-fe70-4ae5-b8b7-25f9adc34f3f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000240435,
      "cost_per_success_usd": 8.0145e-05,
      "octane_score": 22.00178842514609,
      "ci_low": 15.072680766111421,
      "ci_high": 26.5269199575769,
      "ci_metric": "octane_score",
      "median_latency_ms": 1866.0203339532018,
      "p95_latency_ms": 4394.417370483279,
      "input_tokens": 171,
      "output_tokens": 882,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "18f22b4a-6dca-4c1b-8c46-b2f5d1c0e010",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0028320000000000003,
      "cost_per_success_usd": 0.0009440000000000001,
      "octane_score": 2.495762711864406,
      "ci_low": 2.3548163548163545,
      "ci_high": 2.6710334788937407,
      "ci_metric": "octane_score",
      "median_latency_ms": 1585.8541671186686,
      "p95_latency_ms": 3278.5228668712075,
      "input_tokens": 447,
      "output_tokens": 119,
      "reasoning_tokens": 204,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d5548b19-d251-4f4f-9f2c-38d62f031ba4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0028380000000000002,
      "cost_per_success_usd": 0.0009460000000000001,
      "octane_score": 2.2783650458069062,
      "ci_low": 2.1373144399460196,
      "ci_high": 2.452217125382263,
      "ci_metric": "octane_score",
      "median_latency_ms": 3167.6259580999613,
      "p95_latency_ms": 3207.646034564823,
      "input_tokens": 462,
      "output_tokens": 19,
      "reasoning_tokens": 300,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3db5f007-fc7e-4e91-9ded-3b622b77c554",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003468,
      "cost_per_success_usd": 0.0011560000000000001,
      "octane_score": 1.5919838523644751,
      "ci_low": 1.1611785095320624,
      "ci_high": 2.0233160621761654,
      "ci_metric": "octane_score",
      "median_latency_ms": 4021.514290943742,
      "p95_latency_ms": 4679.460867214948,
      "input_tokens": 480,
      "output_tokens": 14,
      "reasoning_tokens": 404,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ddfdaec6-6cb3-4719-89e3-60bafe8cb2ac",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002688,
      "cost_per_success_usd": 0.000896,
      "octane_score": 3.09561011904762,
      "ci_low": 2.961172161172162,
      "ci_high": 3.233716475095786,
      "ci_metric": "octane_score",
      "median_latency_ms": 2999.853542074561,
      "p95_latency_ms": 3195.2753421850502,
      "input_tokens": 351,
      "output_tokens": 75,
      "reasoning_tokens": 256,
      "undefined_reason": null
    },
    {
      "snapshot_id": "979175e3-f3c7-4ab5-a156-48166fc0db54",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0022919999999999998,
      "cost_per_success_usd": 0.0007639999999999999,
      "octane_score": 2.3080279232111693,
      "ci_low": 1.581151832460733,
      "ci_high": 2.9393939393939394,
      "ci_metric": "octane_score",
      "median_latency_ms": 2029.3533746153116,
      "p95_latency_ms": 2928.4258491359647,
      "input_tokens": 306,
      "output_tokens": 31,
      "reasoning_tokens": 249,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b35fcadc-56a4-4834-91b3-102bc09bb7f8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003282,
      "cost_per_success_usd": 0.0010940000000000001,
      "octane_score": 1.9159049360146252,
      "ci_low": 1.8483245149911818,
      "ci_high": 1.9848484848484849,
      "ci_metric": "octane_score",
      "median_latency_ms": 3033.6594162508845,
      "p95_latency_ms": 3165.0026788935065,
      "input_tokens": 648,
      "output_tokens": 26,
      "reasoning_tokens": 305,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3441eac3-0fad-4921-884d-6d6f5d03b4b2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002268,
      "cost_per_success_usd": 0.000756,
      "octane_score": 2.335537918871252,
      "ci_low": 2.3083333333333336,
      "ci_high": 2.3567639257294437,
      "ci_metric": "octane_score",
      "median_latency_ms": 2227.809749543667,
      "p95_latency_ms": 3171.090862434357,
      "input_tokens": 447,
      "output_tokens": 39,
      "reasoning_tokens": 190,
      "undefined_reason": null
    },
    {
      "snapshot_id": "35f93c22-3f01-44e2-b611-a93eca065e80",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.005262,
      "cost_per_success_usd": 0.0017540000000000001,
      "octane_score": 3.8721018624097296,
      "ci_low": 3.720171513795674,
      "ci_high": 4.035917901938427,
      "ci_metric": "octane_score",
      "median_latency_ms": 4184.555999934673,
      "p95_latency_ms": 5516.076675336807,
      "input_tokens": 708,
      "output_tokens": 127,
      "reasoning_tokens": 514,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f2ebc640-3b50-49a2-a73c-e7b467fcecb8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00486,
      "cost_per_success_usd": 0.00162,
      "octane_score": 2.4220164609053505,
      "ci_low": 2.1216702663786897,
      "ci_high": 2.8939169139465872,
      "ci_metric": "octane_score",
      "median_latency_ms": 1713.1358329206705,
      "p95_latency_ms": 3987.329746317118,
      "input_tokens": 384,
      "output_tokens": 209,
      "reasoning_tokens": 473,
      "undefined_reason": null
    },
    {
      "snapshot_id": "90d094b2-7cc3-43be-9415-7ea120e77e0b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0011969,
      "cost_per_success_usd": 0.0003989666666666667,
      "octane_score": 5.253571726961317,
      "ci_low": 4.439737343783097,
      "ci_high": 6.515387006527821,
      "ci_metric": "octane_score",
      "median_latency_ms": 2026.95304248482,
      "p95_latency_ms": 3749.155017640441,
      "input_tokens": 626,
      "output_tokens": 131,
      "reasoning_tokens": 155,
      "undefined_reason": null
    },
    {
      "snapshot_id": "177e06c2-d497-4302-95c8-e115cba83759",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0025281,
      "cost_per_success_usd": 0.0008427,
      "octane_score": 8.059412206795617,
      "ci_low": 7.7946173977579,
      "ci_high": 8.393613160759646,
      "ci_metric": "octane_score",
      "median_latency_ms": 3950.603582896292,
      "p95_latency_ms": 4087.9574957303703,
      "input_tokens": 690,
      "output_tokens": 478,
      "reasoning_tokens": 215,
      "undefined_reason": null
    },
    {
      "snapshot_id": "46a18c00-3690-4793-ae4f-8b45f6f1341b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0013626500000000002,
      "cost_per_success_usd": 0.00045421666666666677,
      "octane_score": 6.106483689869005,
      "ci_low": 5.606198416708776,
      "ci_high": 6.573722010984367,
      "ci_metric": "octane_score",
      "median_latency_ms": 3310.5639591813087,
      "p95_latency_ms": 25642.948634177443,
      "input_tokens": 333,
      "output_tokens": 267,
      "reasoning_tokens": 112,
      "undefined_reason": null
    },
    {
      "snapshot_id": "665caa5c-bbce-4d59-a133-7b191950166b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0021271500000000004,
      "cost_per_success_usd": 0.0007090500000000001,
      "octane_score": 5.533695320029147,
      "ci_low": 5.279587952077035,
      "ci_high": 5.814350374712474,
      "ci_metric": "octane_score",
      "median_latency_ms": 2908.269790932536,
      "p95_latency_ms": 4832.48460367322,
      "input_tokens": 375,
      "output_tokens": 345,
      "reasoning_tokens": 267,
      "undefined_reason": null
    },
    {
      "snapshot_id": "396bfd00-c5a5-4246-8336-54494c556585",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0016513,
      "cost_per_success_usd": 0.0005504333333333334,
      "octane_score": 4.28026403439714,
      "ci_low": 3.545288436439572,
      "ci_high": 6.086282614311546,
      "ci_metric": "octane_score",
      "median_latency_ms": 3874.235125258565,
      "p95_latency_ms": 4617.45238751173,
      "input_tokens": 434,
      "output_tokens": 399,
      "reasoning_tokens": 56,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c84f925b-443d-4ae8-83e8-d640036e4253",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00124035,
      "cost_per_success_usd": 0.00041345000000000005,
      "octane_score": 5.213044705123554,
      "ci_low": 5.004147083218138,
      "ci_high": 5.423618192943096,
      "ci_metric": "octane_score",
      "median_latency_ms": 4451.5634579584,
      "p95_latency_ms": 4869.5956830866635,
      "input_tokens": 431,
      "output_tokens": 90,
      "reasoning_tokens": 237,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4723b49a-d293-498b-9834-12008142b279",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.00132555,
      "cost_per_success_usd": 0.00044185,
      "octane_score": 4.165063558522878,
      "ci_low": 3.0763455299578686,
      "ci_high": 5.289934148635935,
      "ci_metric": "octane_score",
      "median_latency_ms": 5047.999708913267,
      "p95_latency_ms": 8864.200858026743,
      "input_tokens": 471,
      "output_tokens": 0,
      "reasoning_tokens": 348,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3773dbe6-6df3-4d59-b52d-9b86d7907afe",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0008129000000000001,
      "cost_per_success_usd": 0.0002709666666666667,
      "octane_score": 6.516176651494649,
      "ci_low": 5.963075537543623,
      "ci_high": 6.883150419625565,
      "ci_metric": "octane_score",
      "median_latency_ms": 2192.8520826622844,
      "p95_latency_ms": 6255.392383784055,
      "input_tokens": 434,
      "output_tokens": 40,
      "reasoning_tokens": 153,
      "undefined_reason": null
    },
    {
      "snapshot_id": "498de9b3-0688-48c4-8792-6e1a7ea22518",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0012884500000000002,
      "cost_per_success_usd": 0.0004294833333333334,
      "octane_score": 4.105708409329038,
      "ci_low": 2.8958408246434133,
      "ci_high": 5.09677378573534,
      "ci_metric": "octane_score",
      "median_latency_ms": 2216.010667383671,
      "p95_latency_ms": 9963.92406625673,
      "input_tokens": 289,
      "output_tokens": 30,
      "reasoning_tokens": 332,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2e1a8f8f-a378-4507-882a-091f1b631c27",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002922,
      "cost_per_success_usd": 0.0009739999999999999,
      "octane_score": 2.847707049965778,
      "ci_low": 2.511607142857143,
      "ci_high": 3.16962962962963,
      "ci_metric": "octane_score",
      "median_latency_ms": 3249.0455424413085,
      "p95_latency_ms": 4223.518778942525,
      "input_tokens": 321,
      "output_tokens": 81,
      "reasoning_tokens": 299,
      "undefined_reason": null
    },
    {
      "snapshot_id": "eca7a5c5-9f07-4e58-a8f8-98ec130f4204",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.004788,
      "cost_per_success_usd": 0.001596,
      "octane_score": 4.255430242272348,
      "ci_low": 3.798324022346369,
      "ci_high": 4.995827196858125,
      "ci_metric": "octane_score",
      "median_latency_ms": 6148.199125193059,
      "p95_latency_ms": 6238.949912041426,
      "input_tokens": 678,
      "output_tokens": 24,
      "reasoning_tokens": 548,
      "undefined_reason": null
    },
    {
      "snapshot_id": "7480b7a1-339e-48fa-a40e-6369a9ea5cbc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0031079999999999997,
      "cost_per_success_usd": 0.0010359999999999998,
      "octane_score": 2.0231660231660236,
      "ci_low": 1.8614564831261107,
      "ci_high": 2.414746543778802,
      "ci_metric": "octane_score",
      "median_latency_ms": 2890.1257505640388,
      "p95_latency_ms": 3716.5705495513976,
      "input_tokens": 618,
      "output_tokens": 21,
      "reasoning_tokens": 291,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3a9fc542-5561-40fe-b885-e10bd7ba4818",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0027600000000000003,
      "cost_per_success_usd": 0.0009200000000000001,
      "octane_score": 1.9166666666666663,
      "ci_low": 1.3130434782608693,
      "ci_high": 2.4287949921752743,
      "ci_metric": "octane_score",
      "median_latency_ms": 4053.6100836470723,
      "p95_latency_ms": 5162.439521588384,
      "input_tokens": 285,
      "output_tokens": 30,
      "reasoning_tokens": 335,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e8b5417f-f7cd-47ef-b611-d6e97d72ecb5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.003924,
      "cost_per_success_usd": 0.0013080000000000001,
      "octane_score": 2.999745158002039,
      "ci_low": 2.7946894262683735,
      "ci_high": 3.151008492569002,
      "ci_metric": "octane_score",
      "median_latency_ms": 5017.2891253605485,
      "p95_latency_ms": 5369.130775425583,
      "input_tokens": 363,
      "output_tokens": 27,
      "reasoning_tokens": 506,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bbfd2e4d-3c92-4fb4-861f-977714eb6e9d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.002706,
      "cost_per_success_usd": 0.000902,
      "octane_score": 2.6119733924611968,
      "ci_low": 2.465867828612927,
      "ci_high": 2.7901098901098904,
      "ci_metric": "octane_score",
      "median_latency_ms": 3638.7178748846054,
      "p95_latency_ms": 4177.653737459332,
      "input_tokens": 426,
      "output_tokens": 138,
      "reasoning_tokens": 171,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4843f629-d39d-4e58-af84-8ed99547a38f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0027,
      "cost_per_success_usd": 0.0009000000000000001,
      "octane_score": 2.394814814814815,
      "ci_low": 2.231193926846101,
      "ci_high": 2.551044083526682,
      "ci_metric": "octane_score",
      "median_latency_ms": 3525.6714578717947,
      "p95_latency_ms": 4675.98133366555,
      "input_tokens": 423,
      "output_tokens": 21,
      "reasoning_tokens": 288,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2c249f69-4b32-45c9-a4f8-b851d20b18c6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0032700000000000003,
      "cost_per_success_usd": 0.00109,
      "octane_score": 1.6883792048929662,
      "ci_low": 1.259398496240601,
      "ci_high": 2.1495412844036696,
      "ci_metric": "octane_score",
      "median_latency_ms": 3736.403833143413,
      "p95_latency_ms": 3846.4277083054185,
      "input_tokens": 459,
      "output_tokens": 24,
      "reasoning_tokens": 368,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c74ec136-e7bc-4301-aa58-c378f2d1a99f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.002298,
      "cost_per_success_usd": 0.0007660000000000001,
      "octane_score": 2.305047867711053,
      "ci_low": 2.26073883161512,
      "ci_high": 2.3516532618409296,
      "ci_metric": "octane_score",
      "median_latency_ms": 3400.7126251235604,
      "p95_latency_ms": 3694.3311374634504,
      "input_tokens": 426,
      "output_tokens": 17,
      "reasoning_tokens": 224,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9147673c-045f-4acf-af4a-436c9e596647",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.032076,
      "cost_per_success_usd": 0.010692,
      "octane_score": 0.36697219104626516,
      "ci_low": 0.2503557396198364,
      "ci_high": 1.1974327628361858,
      "ci_metric": "octane_score",
      "median_latency_ms": 60363.7003749609,
      "p95_latency_ms": 73113.04247463122,
      "input_tokens": 318,
      "output_tokens": 12,
      "reasoning_tokens": 10627,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f51337af-b83e-4449-90cc-ed1a2f35948b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 2,
      "content_successes": 3,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.003873,
      "cost_per_success_usd": 0.0019365,
      "octane_score": 1.0823650916602119,
      "ci_low": 0.5333333333333334,
      "ci_high": 1.9071883530482256,
      "ci_metric": "octane_score",
      "median_latency_ms": 5568.383709527552,
      "p95_latency_ms": 7342.8033840842545,
      "input_tokens": 564,
      "output_tokens": 342,
      "reasoning_tokens": 855,
      "undefined_reason": null
    },
    {
      "snapshot_id": "35bd8ac9-46c4-4e59-9a50-f7c551b6716c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0060855,
      "cost_per_success_usd": 0.0020285,
      "octane_score": 3.3481225864760495,
      "ci_low": 3.226614924407794,
      "ci_high": 3.491144128263242,
      "ci_metric": "octane_score",
      "median_latency_ms": 9286.10166721046,
      "p95_latency_ms": 9469.307616259903,
      "input_tokens": 627,
      "output_tokens": 9,
      "reasoning_tokens": 1915,
      "undefined_reason": null
    },
    {
      "snapshot_id": "168b30f1-a248-401f-b796-550b21233e60",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0010395,
      "cost_per_success_usd": 0.0003465,
      "octane_score": 5.095719095719096,
      "ci_low": 5.019551740581783,
      "ci_high": 5.173216885007279,
      "ci_metric": "octane_score",
      "median_latency_ms": 2005.3801676258445,
      "p95_latency_ms": 2351.645866408944,
      "input_tokens": 381,
      "output_tokens": 3,
      "reasoning_tokens": 280,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8ff4b7a5-3f5b-4d29-aa9f-1f58f469bdde",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.004656,
      "cost_per_success_usd": 0.0015520000000000002,
      "octane_score": 1.7871563573883162,
      "ci_low": 1.5663466967814794,
      "ci_high": 2.138524800822411,
      "ci_metric": "octane_score",
      "median_latency_ms": 7732.0093754678965,
      "p95_latency_ms": 8316.571337543428,
      "input_tokens": 276,
      "output_tokens": 69,
      "reasoning_tokens": 1437,
      "undefined_reason": null
    },
    {
      "snapshot_id": "18aa5694-34b0-4767-915c-aa10981b73b0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009734999999999999,
      "cost_per_success_usd": 0.0032449999999999996,
      "octane_score": 0.5671289162814588,
      "ci_low": 0.33459879094384126,
      "ci_high": 1.663954189270645,
      "ci_metric": "octane_score",
      "median_latency_ms": 15732.425124384463,
      "p95_latency_ms": 23212.21798751503,
      "input_tokens": 408,
      "output_tokens": 9,
      "reasoning_tokens": 3168,
      "undefined_reason": null
    },
    {
      "snapshot_id": "77a65d70-712b-4fad-86e4-7a6a52d40811",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0022455,
      "cost_per_success_usd": 0.0007485,
      "octane_score": 2.8795368514807396,
      "ci_low": 2.4363577144984188,
      "ci_high": 3.3969004465458372,
      "ci_metric": "octane_score",
      "median_latency_ms": 3716.608707793057,
      "p95_latency_ms": 4552.5655704550445,
      "input_tokens": 369,
      "output_tokens": 6,
      "reasoning_tokens": 681,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6173f0f4-3dd8-4364-a9cd-fb42ed08dac3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0007335,
      "cost_per_success_usd": 0.0002445,
      "octane_score": 7.2119972733469675,
      "ci_low": 5.064989517819706,
      "ci_high": 9.258762117822522,
      "ci_metric": "octane_score",
      "median_latency_ms": 1596.7317912727594,
      "p95_latency_ms": 1609.8401415161788,
      "input_tokens": 225,
      "output_tokens": 15,
      "reasoning_tokens": 192,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fbea94b4-3ca6-457b-8790-5838b52bdaa3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.005367,
      "cost_per_success_usd": 0.001789,
      "octane_score": 1.3169368362213525,
      "ci_low": 1.2568943179714975,
      "ci_high": 1.4152731326644372,
      "ci_metric": "octane_score",
      "median_latency_ms": 8579.120167531073,
      "p95_latency_ms": 8876.985104568303,
      "input_tokens": 372,
      "output_tokens": 93,
      "reasoning_tokens": 1634,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c4b24ffd-e723-4dc9-a6cc-a9614467de94",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-missing-weight-refusal",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.009599,
      "cost_per_success_usd": 0.0031996666666666666,
      "octane_score": 0.5751640795916242,
      "ci_low": 0.3952237239207453,
      "ci_high": 0.8400180261378999,
      "ci_metric": "octane_score",
      "median_latency_ms": 19576.72058325261,
      "p95_latency_ms": 25171.54980795458,
      "input_tokens": 453,
      "output_tokens": 9,
      "reasoning_tokens": 3740,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d2fd6527-9b77-498c-ad4f-9c6a0dae3f7e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-max-dose-cap",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0019345,
      "cost_per_success_usd": 0.0006448333333333333,
      "octane_score": 3.3424657534246585,
      "ci_low": 2.9694957386363647,
      "ci_high": 3.859807460890493,
      "ci_metric": "octane_score",
      "median_latency_ms": 5222.760500386357,
      "p95_latency_ms": 5256.006911955774,
      "input_tokens": 414,
      "output_tokens": 6,
      "reasoning_tokens": 685,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cd566156-9772-4d94-a5a6-0e12c7677487",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-interaction-extraction",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0038885,
      "cost_per_success_usd": 0.0012961666666666666,
      "octane_score": 1.8176674810338174,
      "ci_low": 1.509389671361502,
      "ci_high": 2.3396226415094334,
      "ci_metric": "octane_score",
      "median_latency_ms": 11307.713250629604,
      "p95_latency_ms": 12929.584237094969,
      "input_tokens": 417,
      "output_tokens": 93,
      "reasoning_tokens": 1379,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fb17d3da-b127-4ae5-8fca-33d1c9c2800f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-infusion-rate-units",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0217665,
      "cost_per_success_usd": 0.007255500000000001,
      "octane_score": 0.5407851514942688,
      "ci_low": 0.40281984155501954,
      "ci_high": 0.6980224478888296,
      "ci_metric": "octane_score",
      "median_latency_ms": 37070.39791718125,
      "p95_latency_ms": 52710.41495408862,
      "input_tokens": 363,
      "output_tokens": 12,
      "reasoning_tokens": 8622,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cce06a1b-fb7c-4a6f-9ccd-3995c54374e9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-icd10-lookup",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000595,
      "cost_per_success_usd": 0.00019833333333333335,
      "octane_score": 8.890756302521007,
      "ci_low": 6.142372881355931,
      "ci_high": 10.796521739130435,
      "ci_metric": "octane_score",
      "median_latency_ms": 2457.864750176668,
      "p95_latency_ms": 2566.170750092715,
      "input_tokens": 270,
      "output_tokens": 15,
      "reasoning_tokens": 169,
      "undefined_reason": null
    },
    {
      "snapshot_id": "27e21e2f-6010-4fcc-99a4-87bd27a24fdf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-conditional-interaction",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0023295,
      "cost_per_success_usd": 0.0007765,
      "octane_score": 2.699291693496459,
      "ci_low": 2.6820217530390273,
      "ci_high": 2.708010335917313,
      "ci_metric": "octane_score",
      "median_latency_ms": 6279.076457954943,
      "p95_latency_ms": 6357.0466462522745,
      "input_tokens": 609,
      "output_tokens": 6,
      "reasoning_tokens": 804,
      "undefined_reason": null
    },
    {
      "snapshot_id": "de7128fc-979c-4ee5-a9e9-17c73a62ea0e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-reference-range-interpretation",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.000723,
      "cost_per_success_usd": 0.000241,
      "octane_score": 7.326417704011065,
      "ci_low": 7.232365145228217,
      "ci_high": 7.37344398340249,
      "ci_metric": "octane_score",
      "median_latency_ms": 3147.0387084409595,
      "p95_latency_ms": 3513.470033276826,
      "input_tokens": 426,
      "output_tokens": 3,
      "reasoning_tokens": 201,
      "undefined_reason": null
    },
    {
      "snapshot_id": "96acff3c-3654-4d82-bc73-ef824a799b49",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-renal-dose-adjustment",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 1.0,
      "total_cost_usd": 0.0070209999999999995,
      "cost_per_success_usd": 0.002340333333333333,
      "octane_score": 2.9020082609314914,
      "ci_low": 2.3818532142231565,
      "ci_high": 4.4404489484581005,
      "ci_metric": "octane_score",
      "median_latency_ms": 15108.298417180777,
      "p95_latency_ms": 15518.587755318731,
      "input_tokens": 672,
      "output_tokens": 9,
      "reasoning_tokens": 2665,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6021c338-3f34-4250-b8b0-159f1ba0827f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "task",
      "vertical": "medical",
      "task_id": "medical-paediatric-dose-volume",
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 3,
      "successes": 3,
      "content_successes": null,
      "success_rate": 1.0,
      "total_cost_usd": 0.0020930000000000002,
      "cost_per_success_usd": 0.0006976666666666667,
      "octane_score": 3.975633062589585,
      "ci_low": 3.805542283803154,
      "ci_high": 4.13536701620591,
      "ci_metric": "octane_score",
      "median_latency_ms": 6123.863792046905,
      "p95_latency_ms": 6247.904453985393,
      "input_tokens": 321,
      "output_tokens": 69,
      "reasoning_tokens": 704,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c938985f-9e80-4d51-a247-cce49e800096",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.003645,
      "cost_per_success_usd": 0.00019544444444444446,
      "octane_score": 19.678226264923254,
      "ci_low": 10.640681203348963,
      "ci_high": 36.732768811067324,
      "ci_metric": "octane_score",
      "median_latency_ms": 628.5590003244579,
      "p95_latency_ms": 1235.462213912976,
      "input_tokens": 1785,
      "output_tokens": 372,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b122a095-8376-40b7-9713-6ada412cd96e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 15,
      "content_successes": 18,
      "success_rate": 0.8333333333333334,
      "total_cost_usd": 0.052950000000000004,
      "cost_per_success_usd": 0.0030634375000000006,
      "octane_score": 1.2554524125267776,
      "ci_low": 0.4846545169258549,
      "ci_high": 2.0641089899666167,
      "ci_metric": "octane_score",
      "median_latency_ms": 2589.994396083057,
      "p95_latency_ms": 6158.897547982631,
      "input_tokens": 1785,
      "output_tokens": 1761,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ed03a951-5954-4406-ad22-5220ba1f4885",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.01584,
      "cost_per_success_usd": 0.0008588333333333334,
      "octane_score": 4.478168057442267,
      "ci_low": 3.517244671241039,
      "ci_high": 5.803843091137143,
      "ci_metric": "octane_score",
      "median_latency_ms": 1451.385228894651,
      "p95_latency_ms": 3299.9458919744934,
      "input_tokens": 1785,
      "output_tokens": 699,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "70dec306-d19b-4e86-9417-5870eb59d8bd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 0.8333333333333334,
      "total_cost_usd": 0.00055658,
      "cost_per_success_usd": 3.947142857142857e-05,
      "octane_score": 97.43756786102065,
      "ci_low": 40.73600255850313,
      "ci_high": 234.3663013643659,
      "ci_metric": "octane_score",
      "median_latency_ms": 1235.7106050476432,
      "p95_latency_ms": 2318.5330124106254,
      "input_tokens": 1686,
      "output_tokens": 333,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f612f129-4992-4b13-be05-f93e65c24ac9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.00078568,
      "cost_per_success_usd": 4.36074074074074e-05,
      "octane_score": 88.19602514013933,
      "ci_low": 47.01195781461061,
      "ci_high": 159.26002189205136,
      "ci_metric": "octane_score",
      "median_latency_ms": 3494.089020881802,
      "p95_latency_ms": 35514.49573948046,
      "input_tokens": 2369,
      "output_tokens": 829,
      "reasoning_tokens": 2483,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f52acc7a-e794-4331-9aa1-640fed02c872",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 17,
      "content_successes": 17,
      "success_rate": 0.9444444444444444,
      "total_cost_usd": 0.00365748,
      "cost_per_success_usd": 0.00022072411764705887,
      "octane_score": 17.42446652861837,
      "ci_low": 10.740462338385868,
      "ci_high": 28.26664558007454,
      "ci_metric": "octane_score",
      "median_latency_ms": 3567.208082880825,
      "p95_latency_ms": 7575.932873040438,
      "input_tokens": 2684,
      "output_tokens": 278,
      "reasoning_tokens": 2584,
      "undefined_reason": null
    },
    {
      "snapshot_id": "df3fdd7e-5f86-45c8-a50c-bca607c9bea5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.0015978,
      "cost_per_success_usd": 8.269999999999999e-05,
      "octane_score": 46.50544135429264,
      "ci_low": 22.019310273187575,
      "ci_high": 96.77339790678425,
      "ci_metric": "octane_score",
      "median_latency_ms": 714.6785627119243,
      "p95_latency_ms": 833.820313727483,
      "input_tokens": 1626,
      "output_tokens": 444,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d42f8877-7d59-43b4-a970-23078511e02b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.078564,
      "cost_per_success_usd": 0.0043046111111111115,
      "octane_score": 0.8934605010131256,
      "ci_low": 0.4934290743617396,
      "ci_high": 1.612358233987502,
      "ci_metric": "octane_score",
      "median_latency_ms": 2901.205896399915,
      "p95_latency_ms": 7638.765741325909,
      "input_tokens": 1636,
      "output_tokens": 406,
      "reasoning_tokens": 9742,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ae92b7e1-baf5-408d-a469-030540d7a7dd",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.012350999999999999,
      "cost_per_success_usd": 0.0006653888888888889,
      "octane_score": 5.780078483760542,
      "ci_low": 3.4968179735929708,
      "ci_high": 9.933811056400362,
      "ci_metric": "octane_score",
      "median_latency_ms": 3140.2420420199633,
      "p95_latency_ms": 5478.046681778503,
      "input_tokens": 1626,
      "output_tokens": 389,
      "reasoning_tokens": 5873,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a256bb16-7e6c-4e11-8e54-688e3ef8aa6d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.005853860000000001,
      "cost_per_success_usd": 0.00029787925925925927,
      "octane_score": 12.911271531841141,
      "ci_low": 6.982504321586494,
      "ci_high": 21.673032133495873,
      "ci_metric": "octane_score",
      "median_latency_ms": 11955.239041242748,
      "p95_latency_ms": 36267.42349588308,
      "input_tokens": 1551,
      "output_tokens": 1933,
      "reasoning_tokens": 12489,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8d45ff88-97c6-40c6-99df-7a50fed752ff",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.023450159999999998,
      "cost_per_success_usd": 0.001305542222222222,
      "octane_score": 2.945902426237546,
      "ci_low": 1.5119741585353883,
      "ci_high": 5.646832113943426,
      "ci_metric": "octane_score",
      "median_latency_ms": 13972.614874131978,
      "p95_latency_ms": 98020.8793676458,
      "input_tokens": 1586,
      "output_tokens": 541,
      "reasoning_tokens": 11287,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4cbcfdb5-06ec-4b77-83f9-779b3d7aaa04",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.0193342692,
      "cost_per_success_usd": 0.0009451763111111111,
      "octane_score": 4.069082090598313,
      "ci_low": 2.0229687940619443,
      "ci_high": 6.8169704303407554,
      "ci_metric": "octane_score",
      "median_latency_ms": 3140.968875028193,
      "p95_latency_ms": 11705.196029087527,
      "input_tokens": 1677,
      "output_tokens": 379,
      "reasoning_tokens": 5750,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4835229b-0c12-4e1d-b773-38241863b833",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.017453800000000002,
      "cost_per_success_usd": 0.0008638251851851853,
      "octane_score": 4.452289729403412,
      "ci_low": 1.9609312667428853,
      "ci_high": 8.972985723184445,
      "ci_metric": "octane_score",
      "median_latency_ms": 3604.991937521845,
      "p95_latency_ms": 11202.020539389918,
      "input_tokens": 1677,
      "output_tokens": 297,
      "reasoning_tokens": 3302,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3d893c34-5fd7-4d8e-be12-9ae2d0719bc4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.00245925,
      "cost_per_success_usd": 0.00013291666666666665,
      "octane_score": 28.935423197492174,
      "ci_low": 16.021942355281368,
      "ci_high": 55.351836328734066,
      "ci_metric": "octane_score",
      "median_latency_ms": 735.6726252473891,
      "p95_latency_ms": 916.5647482965137,
      "input_tokens": 1563,
      "output_tokens": 286,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e775d284-8665-4590-ac00-ce4a4fcaf33e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.0013386000000000001,
      "cost_per_success_usd": 7.335555555555557e-05,
      "octane_score": 52.42956679794002,
      "ci_low": 33.27665411957781,
      "ci_high": 81.36657351619499,
      "ci_metric": "octane_score",
      "median_latency_ms": 1881.066916976124,
      "p95_latency_ms": 2238.677302422002,
      "input_tokens": 1563,
      "output_tokens": 333,
      "reasoning_tokens": 522,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0340c980-aa4d-4ecb-bd07-42886af748aa",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.011025,
      "cost_per_success_usd": 0.0006092111111111111,
      "octane_score": 6.313082492841381,
      "ci_low": 4.057793198638632,
      "ci_high": 10.130192809444102,
      "ci_metric": "octane_score",
      "median_latency_ms": 3360.2302707731724,
      "p95_latency_ms": 3976.94269567728,
      "input_tokens": 34881,
      "output_tokens": 1303,
      "reasoning_tokens": 2071,
      "undefined_reason": null
    },
    {
      "snapshot_id": "584471d1-d917-4d36-8618-21b7ee9a9601",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.006096,
      "cost_per_success_usd": 0.0003303333333333333,
      "octane_score": 11.642785065590317,
      "ci_low": 6.357707913778534,
      "ci_high": 22.818710760428694,
      "ci_metric": "octane_score",
      "median_latency_ms": 1209.6918541938066,
      "p95_latency_ms": 1781.4142496790719,
      "input_tokens": 1563,
      "output_tokens": 297,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bae30cd1-281b-4446-8f20-d1a339c66b96",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.135765,
      "cost_per_success_usd": 0.007456,
      "octane_score": 0.5158261802575108,
      "ci_low": 0.2906599547089579,
      "ci_high": 0.9433004334209294,
      "ci_metric": "octane_score",
      "median_latency_ms": 2855.702292174101,
      "p95_latency_ms": 4121.289387717843,
      "input_tokens": 33549,
      "output_tokens": 1163,
      "reasoning_tokens": 69,
      "undefined_reason": null
    },
    {
      "snapshot_id": "81af13a0-c3a5-4ac7-ab79-f55ac3da9085",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.009054,
      "cost_per_success_usd": 0.0005086666666666667,
      "octane_score": 7.560943643512452,
      "ci_low": 5.382735830023202,
      "ci_high": 10.89903509466813,
      "ci_metric": "octane_score",
      "median_latency_ms": 1006.155916955322,
      "p95_latency_ms": 2256.2364003155376,
      "input_tokens": 1563,
      "output_tokens": 311,
      "reasoning_tokens": 183,
      "undefined_reason": null
    },
    {
      "snapshot_id": "98c21d75-2a7b-4c15-846d-2d735c442cc9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.040256,
      "cost_per_success_usd": 0.0022843888888888888,
      "octane_score": 1.6836012548943315,
      "ci_low": 1.1343161326556004,
      "ci_high": 2.5656365099235314,
      "ci_metric": "octane_score",
      "median_latency_ms": 2396.256832871586,
      "p95_latency_ms": 3431.416408205405,
      "input_tokens": 34040,
      "output_tokens": 1254,
      "reasoning_tokens": 1045,
      "undefined_reason": null
    },
    {
      "snapshot_id": "511483be-513a-4d46-89f3-180072012172",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.00068612,
      "cost_per_success_usd": 3.846000000000001e-05,
      "octane_score": 100.0,
      "ci_low": 2.332529595588235e-05,
      "ci_high": 6.410975146198829e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 2956.384957768023,
      "p95_latency_ms": 12512.280109617854,
      "input_tokens": 2618,
      "output_tokens": 1464,
      "reasoning_tokens": 2110,
      "undefined_reason": null
    },
    {
      "snapshot_id": "1111c6ea-209d-41cb-9886-a62fecda8216",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.0265324,
      "cost_per_success_usd": 0.0014871629629629626,
      "octane_score": 2.5861321830778974,
      "ci_low": 1.8006720254145925,
      "ci_high": 3.707893877083038,
      "ci_metric": "octane_score",
      "median_latency_ms": 2852.635000832379,
      "p95_latency_ms": 3596.310777636245,
      "input_tokens": 5181,
      "output_tokens": 261,
      "reasoning_tokens": 3377,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d8127402-549a-4714-bc12-a8b468408317",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.039636,
      "cost_per_success_usd": 0.0022203333333333333,
      "octane_score": 1.732172346494521,
      "ci_low": 1.0581324347221683,
      "ci_high": 2.68961801797539,
      "ci_metric": "octane_score",
      "median_latency_ms": 4221.483458764851,
      "p95_latency_ms": 9294.11474168299,
      "input_tokens": 5181,
      "output_tokens": 261,
      "reasoning_tokens": 5450,
      "undefined_reason": null
    },
    {
      "snapshot_id": "90fe2125-9435-43f1-955e-cfddffab9e6e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.02141447,
      "cost_per_success_usd": 0.001206689814814815,
      "octane_score": 3.1872316752671264,
      "ci_low": 1.8268437985107109,
      "ci_high": 5.556754938450337,
      "ci_metric": "octane_score",
      "median_latency_ms": 9748.673292342573,
      "p95_latency_ms": 21384.406751347662,
      "input_tokens": 1613,
      "output_tokens": 700,
      "reasoning_tokens": 8550,
      "undefined_reason": null
    },
    {
      "snapshot_id": "04f7bafb-3087-4577-999a-1e113ac68abb",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.09437999999999999,
      "cost_per_success_usd": 0.004992333333333332,
      "octane_score": 0.77038125125192,
      "ci_low": 0.5472330685362399,
      "ci_high": 1.136937395572264,
      "ci_metric": "octane_score",
      "median_latency_ms": 2343.201062642038,
      "p95_latency_ms": 5363.973078830167,
      "input_tokens": 2988,
      "output_tokens": 1973,
      "reasoning_tokens": 4067,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c533678e-4661-40ed-ba86-6fabbc222b04",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 9,
      "content_successes": 15,
      "success_rate": 0.5,
      "total_cost_usd": 0.0027566,
      "cost_per_success_usd": 0.0004365809523809524,
      "octane_score": 8.809362797495693,
      "ci_low": 1.9633415930358793,
      "ci_high": 23.198891621690848,
      "ci_metric": "octane_score",
      "median_latency_ms": 2372.488208580762,
      "p95_latency_ms": 12776.780949858945,
      "input_tokens": 1623,
      "output_tokens": 3040,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "32f5885d-d873-44ed-9dac-f94f818cc1b3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 29.62262462824583,
      "p95_latency_ms": 53.10479984618717,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "0a11293c-44ac-45f5-91c1-c3bce6af2e99",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 16,
      "content_successes": 18,
      "success_rate": 0.8888888888888888,
      "total_cost_usd": 0.10788,
      "cost_per_success_usd": 0.006387291666666666,
      "octane_score": 0.6021331419811476,
      "ci_low": 0.30727365820509284,
      "ci_high": 1.019352497661592,
      "ci_metric": "octane_score",
      "median_latency_ms": 2491.842937655747,
      "p95_latency_ms": 3795.962940901516,
      "input_tokens": 2478,
      "output_tokens": 631,
      "reasoning_tokens": 1031,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8c6760ab-9afd-4aa0-88cf-c279227f3987",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.005304585,
      "cost_per_success_usd": 0.0004129811111111111,
      "octane_score": 9.312774595555892,
      "ci_low": 3.6384170387572485,
      "ci_high": 16.121172160758377,
      "ci_metric": "octane_score",
      "median_latency_ms": 4557.228582911193,
      "p95_latency_ms": 12236.436962988224,
      "input_tokens": 1797,
      "output_tokens": 19953,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "12badca2-3d4f-432f-8cf0-9204c2851adc",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.0264768,
      "cost_per_success_usd": 0.001500711111111111,
      "octane_score": 2.5627850500503473,
      "ci_low": 1.9635213860035265,
      "ci_high": 3.3125893570259453,
      "ci_metric": "octane_score",
      "median_latency_ms": 2800.076812505722,
      "p95_latency_ms": 6807.628818880765,
      "input_tokens": 2607,
      "output_tokens": 1177,
      "reasoning_tokens": 2386,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d5ddc4be-44f4-478c-a59f-28e67f67baaf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.01200465,
      "cost_per_success_usd": 0.0006943962962962963,
      "octane_score": 5.538624011264782,
      "ci_low": 4.196243876649577,
      "ci_high": 7.353746596206879,
      "ci_metric": "octane_score",
      "median_latency_ms": 2888.254666700959,
      "p95_latency_ms": 11571.968198474488,
      "input_tokens": 2549,
      "output_tokens": 1446,
      "reasoning_tokens": 1947,
      "undefined_reason": null
    },
    {
      "snapshot_id": "49d58ea0-d9db-4040-a89b-66fd80c93632",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.027024,
      "cost_per_success_usd": 0.0015352222222222223,
      "octane_score": 2.505174784685533,
      "ci_low": 2.0842008996169907,
      "ci_high": 3.047706557242995,
      "ci_metric": "octane_score",
      "median_latency_ms": 3688.1106668151915,
      "p95_latency_ms": 11992.21421116963,
      "input_tokens": 2481,
      "output_tokens": 411,
      "reasoning_tokens": 3266,
      "undefined_reason": null
    },
    {
      "snapshot_id": "597339ee-b4b3-4838-b5f6-02628aa7a420",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.079107,
      "cost_per_success_usd": 0.00520275,
      "octane_score": 0.7392244486089089,
      "ci_low": 0.27748874527573936,
      "ci_high": 2.8533660170075383,
      "ci_metric": "octane_score",
      "median_latency_ms": 8509.278708137572,
      "p95_latency_ms": 72830.65629657349,
      "input_tokens": 2226,
      "output_tokens": 311,
      "reasoning_tokens": 25687,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fca0f28f-267c-44c8-966b-48735e26192e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "banking",
      "task_id": null,
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 18,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 1.0,
      "total_cost_usd": 0.0323605,
      "cost_per_success_usd": 0.0019662685185185187,
      "octane_score": 1.9559892068544953,
      "ci_low": 1.2059349806387472,
      "ci_high": 3.6218791644023494,
      "ci_metric": "octane_score",
      "median_latency_ms": 7659.032437950373,
      "p95_latency_ms": 35794.632679130846,
      "input_tokens": 2496,
      "output_tokens": 309,
      "reasoning_tokens": 12136,
      "undefined_reason": null
    },
    {
      "snapshot_id": "65a4171f-f004-4843-8b2c-12f4e5583030",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.023982,
      "cost_per_success_usd": 0.0012486956521739132,
      "octane_score": 9.22159792157703,
      "ci_low": 6.98493686469599,
      "ci_high": 13.163085297407418,
      "ci_metric": "octane_score",
      "median_latency_ms": 2175.5019165575504,
      "p95_latency_ms": 3561.7909152060747,
      "input_tokens": 2352,
      "output_tokens": 4326,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "743ebe70-deab-44f0-ad49-b7d27097316e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.114635,
      "cost_per_success_usd": 0.006011956521739131,
      "octane_score": 1.9153447283145348,
      "ci_low": 1.5637012326068882,
      "ci_high": 2.723843173882425,
      "ci_metric": "octane_score",
      "median_latency_ms": 4231.3127079978585,
      "p95_latency_ms": 6281.365083530545,
      "input_tokens": 2352,
      "output_tokens": 4115,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f21d40b1-2f6e-4d9a-b2ab-ab8d00e5ecec",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.063996,
      "cost_per_success_usd": 0.0033519565217391308,
      "octane_score": 3.4352979091729985,
      "ci_low": 2.6641036416315678,
      "ci_high": 5.763844419997442,
      "ci_metric": "octane_score",
      "median_latency_ms": 2701.2603748589754,
      "p95_latency_ms": 6473.550166934729,
      "input_tokens": 2352,
      "output_tokens": 3796,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b0b251e2-e157-4071-bb7c-18f979523ed3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.0019794,
      "cost_per_success_usd": 0.00010007797101449277,
      "octane_score": 115.0599788748884,
      "ci_low": 78.84691140789688,
      "ci_high": 198.56964787754117,
      "ci_metric": "octane_score",
      "median_latency_ms": 3891.5557507425547,
      "p95_latency_ms": 9335.692624561489,
      "input_tokens": 2214,
      "output_tokens": 3716,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9aa56c42-0ebf-48da-a93d-0c29f1a1b477",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 16,
      "content_successes": 16,
      "success_rate": 0.7619047619047619,
      "total_cost_usd": 0.054426844,
      "cost_per_success_usd": 0.003999011307692308,
      "octane_score": 2.8794540312050594,
      "ci_low": 0.696076845349941,
      "ci_high": 31.321360403900968,
      "ci_metric": "octane_score",
      "median_latency_ms": 25541.83083306998,
      "p95_latency_ms": 237857.34912473708,
      "input_tokens": 2963,
      "output_tokens": 12698,
      "reasoning_tokens": 288413,
      "undefined_reason": null
    },
    {
      "snapshot_id": "af074410-b7be-4f61-96a9-edc16ffe113a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 0.7142857142857143,
      "total_cost_usd": 0.317802735,
      "cost_per_success_usd": 0.021858083617021278,
      "octane_score": 0.526805983201671,
      "ci_low": 0.15522366220978115,
      "ci_high": 5.903014782085349,
      "ci_metric": "octane_score",
      "median_latency_ms": 18496.8149587512,
      "p95_latency_ms": 749616.1389583722,
      "input_tokens": 3345,
      "output_tokens": 5966,
      "reasoning_tokens": 357652,
      "undefined_reason": null
    },
    {
      "snapshot_id": "880f6674-103f-4a1b-8540-5f20324e2a54",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 20,
      "content_successes": 20,
      "success_rate": 0.9523809523809523,
      "total_cost_usd": 0.0188662,
      "cost_per_success_usd": 0.0010452430769230768,
      "octane_score": 11.016546758354334,
      "ci_low": 7.147590413622892,
      "ci_high": 19.761171358516165,
      "ci_metric": "octane_score",
      "median_latency_ms": 1787.4715840443969,
      "p95_latency_ms": 4604.653749614954,
      "input_tokens": 2154,
      "output_tokens": 7288,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "41582c6e-da64-4192-bc2a-abdf705e82ef",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 20,
      "content_successes": 20,
      "success_rate": 0.9523809523809523,
      "total_cost_usd": 0.24909900000000001,
      "cost_per_success_usd": 0.014065315384615383,
      "octane_score": 0.8186783528056744,
      "ci_low": 0.6131000233362299,
      "ci_high": 1.156999388793433,
      "ci_metric": "octane_score",
      "median_latency_ms": 7163.5389579460025,
      "p95_latency_ms": 18717.768625356257,
      "input_tokens": 2171,
      "output_tokens": 2870,
      "reasoning_tokens": 29909,
      "undefined_reason": null
    },
    {
      "snapshot_id": "135907a6-8f95-4d62-b805-7b35120b939e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.037773375,
      "cost_per_success_usd": 0.001984570652173913,
      "octane_score": 5.802247059410886,
      "ci_low": 4.874993559475795,
      "ci_high": 7.755095959630022,
      "ci_metric": "octane_score",
      "median_latency_ms": 6713.720457628369,
      "p95_latency_ms": 11772.113291546702,
      "input_tokens": 2154,
      "output_tokens": 2926,
      "reasoning_tokens": 16789,
      "undefined_reason": null
    },
    {
      "snapshot_id": "21b89472-3df6-4a48-86d2-f68ea5de1e6b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.5714285714285714,
      "total_cost_usd": 0.058620200000000004,
      "cost_per_success_usd": 0.005745287027027027,
      "octane_score": 2.004246119750052,
      "ci_low": 0.6918113094327084,
      "ci_high": 6.022604347415234,
      "ci_metric": "octane_score",
      "median_latency_ms": 58776.84145886451,
      "p95_latency_ms": 319986.97487451136,
      "input_tokens": 1955,
      "output_tokens": 7136,
      "reasoning_tokens": 139167,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0cfecb1c-26fc-4952-9a7a-0334e3b46d9e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.5714285714285714,
      "total_cost_usd": 0.28871147999999996,
      "cost_per_success_usd": 0.030273414857142855,
      "octane_score": 0.3803657197282564,
      "ci_low": 0.1148938530612196,
      "ci_high": 1.9057467019567709,
      "ci_metric": "octane_score",
      "median_latency_ms": 153520.2424582094,
      "p95_latency_ms": 2534352.4756655097,
      "input_tokens": 2113,
      "output_tokens": 4438,
      "reasoning_tokens": 145377,
      "undefined_reason": null
    },
    {
      "snapshot_id": "23904e27-91d6-48cd-8d3e-4c599a8cf04d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 19,
      "content_successes": 19,
      "success_rate": 0.9047619047619048,
      "total_cost_usd": 0.18175775760000001,
      "cost_per_success_usd": 0.011452800078688525,
      "octane_score": 1.0054282927889742,
      "ci_low": 0.48295405465718644,
      "ci_high": 3.656169430336796,
      "ci_metric": "octane_score",
      "median_latency_ms": 10060.66033244133,
      "p95_latency_ms": 155567.7723744884,
      "input_tokens": 2232,
      "output_tokens": 2712,
      "reasoning_tokens": 56847,
      "undefined_reason": null
    },
    {
      "snapshot_id": "761964f2-17e0-417f-93db-b68b06aab8bf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.09883224000000002,
      "cost_per_success_usd": 0.005092013913043479,
      "octane_score": 2.261378194838194,
      "ci_low": 1.5371061873273286,
      "ci_high": 4.025284727256517,
      "ci_metric": "octane_score",
      "median_latency_ms": 11125.712499953806,
      "p95_latency_ms": 46117.9705420509,
      "input_tokens": 2232,
      "output_tokens": 2422,
      "reasoning_tokens": 19595,
      "undefined_reason": null
    },
    {
      "snapshot_id": "dbbc228f-631b-4f1f-a720-c78f511ae570",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.013752,
      "cost_per_success_usd": 0.0007137065217391305,
      "octane_score": 16.13403952469151,
      "ci_low": 12.829409988510399,
      "ci_high": 23.224012251140376,
      "ci_metric": "octane_score",
      "median_latency_ms": 1192.913917824626,
      "p95_latency_ms": 1681.1710419133306,
      "input_tokens": 2106,
      "output_tokens": 2705,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "79ad3f04-4488-4a9c-8829-8116a7eb12ee",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.0088836,
      "cost_per_success_usd": 0.00045383478260869564,
      "octane_score": 25.372601819058108,
      "ci_low": 15.386731696055607,
      "ci_high": 55.71207197470289,
      "ci_metric": "octane_score",
      "median_latency_ms": 2858.719792217016,
      "p95_latency_ms": 10245.120209641755,
      "input_tokens": 2106,
      "output_tokens": 2329,
      "reasoning_tokens": 4723,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6d826089-dffb-4a6c-a40e-ad0e931d5071",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.04139804,
      "cost_per_success_usd": 0.0020787808695652173,
      "octane_score": 5.539289590045929,
      "ci_low": 3.3968048829564985,
      "ci_high": 10.254927034038982,
      "ci_metric": "octane_score",
      "median_latency_ms": 5565.047457814217,
      "p95_latency_ms": 22111.086499877274,
      "input_tokens": 59659,
      "output_tokens": 7091,
      "reasoning_tokens": 17880,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6c25c6e4-38fc-4754-b63f-13007e158c01",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.059092,
      "cost_per_success_usd": 0.0030483768115942026,
      "octane_score": 3.7774100586821073,
      "ci_low": 2.363862167939887,
      "ci_high": 8.926522610750965,
      "ci_metric": "octane_score",
      "median_latency_ms": 3233.7110405787826,
      "p95_latency_ms": 14462.00858335942,
      "input_tokens": 2106,
      "output_tokens": 2326,
      "reasoning_tokens": 3162,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f565390c-b7ed-4d44-a91f-a4a4a60e01b6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.927311,
      "cost_per_success_usd": 0.04803473913043478,
      "octane_score": 0.23972169807149749,
      "ci_low": 0.14794234285164468,
      "ci_high": 0.584329362033647,
      "ci_metric": "octane_score",
      "median_latency_ms": 6967.439957894385,
      "p95_latency_ms": 34853.68979163468,
      "input_tokens": 57970,
      "output_tokens": 7345,
      "reasoning_tokens": 14816,
      "undefined_reason": null
    },
    {
      "snapshot_id": "676f48fb-4a64-4233-8d30-35aa44c20db4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.075156,
      "cost_per_success_usd": 0.003890521739130434,
      "octane_score": 2.959749360851259,
      "ci_low": 1.7938799487919732,
      "ci_high": 6.975267896899537,
      "ci_metric": "octane_score",
      "median_latency_ms": 1587.8317914903164,
      "p95_latency_ms": 9493.000040762126,
      "input_tokens": 2106,
      "output_tokens": 2312,
      "reasoning_tokens": 3600,
      "undefined_reason": null
    },
    {
      "snapshot_id": "615b2087-d0e8-40f0-a103-30822b149b7f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.1873523,
      "cost_per_success_usd": 0.009571227536231885,
      "octane_score": 1.2030817559377114,
      "ci_low": 0.7833123554968344,
      "ci_high": 2.395654711748576,
      "ci_metric": "octane_score",
      "median_latency_ms": 5122.547125443816,
      "p95_latency_ms": 24392.81849935651,
      "input_tokens": 57515,
      "output_tokens": 7459,
      "reasoning_tokens": 14814,
      "undefined_reason": null
    },
    {
      "snapshot_id": "67ea3090-05f5-464f-965f-9b039b5153cf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 20,
      "content_successes": 20,
      "success_rate": 0.9523809523809523,
      "total_cost_usd": 0.0020822,
      "cost_per_success_usd": 0.00011514969230769231,
      "octane_score": 100.0,
      "ci_low": 6.491406505376345e-05,
      "ci_high": 0.00016765402248362116,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 12714.398249983788,
      "p95_latency_ms": 39270.4091668129,
      "input_tokens": 3339,
      "output_tokens": 4604,
      "reasoning_tokens": 7055,
      "undefined_reason": null
    },
    {
      "snapshot_id": "604688f3-472a-4d60-a62d-3e4bda6c9e2a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.1104272,
      "cost_per_success_usd": 0.005782689855072464,
      "octane_score": 1.9912825206540383,
      "ci_low": 1.541257949178346,
      "ci_high": 2.8331465717885194,
      "ci_metric": "octane_score",
      "median_latency_ms": 11029.069290496409,
      "p95_latency_ms": 27201.978833414614,
      "input_tokens": 6375,
      "output_tokens": 2926,
      "reasoning_tokens": 14369,
      "undefined_reason": null
    },
    {
      "snapshot_id": "207d9929-cbe4-47d6-b293-6d5d6b4cf95f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.156006,
      "cost_per_success_usd": 0.00827286956521739,
      "octane_score": 1.3918954166983348,
      "ci_low": 1.0236640317313817,
      "ci_high": 2.0739274194652877,
      "ci_metric": "octane_score",
      "median_latency_ms": 13863.440833054483,
      "p95_latency_ms": 37173.653666861355,
      "input_tokens": 6375,
      "output_tokens": 2429,
      "reasoning_tokens": 22407,
      "undefined_reason": null
    },
    {
      "snapshot_id": "970a20e8-9070-4c69-bd18-29574a019eba",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.16685075,
      "cost_per_success_usd": 0.00898672579710145,
      "octane_score": 1.2813308751985326,
      "ci_low": 0.8758071863929399,
      "ci_high": 2.2292969602955375,
      "ci_metric": "octane_score",
      "median_latency_ms": 44933.581999503076,
      "p95_latency_ms": 222540.29079154134,
      "input_tokens": 2144,
      "output_tokens": 4217,
      "reasoning_tokens": 69554,
      "undefined_reason": null
    },
    {
      "snapshot_id": "537dc2bf-5758-47c3-ab5a-79c53bcfa064",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 1.0,
      "total_cost_usd": 0.2626788,
      "cost_per_success_usd": 0.013492843478260871,
      "octane_score": 0.8534130888957312,
      "ci_low": 0.5578641907058439,
      "ci_high": 1.7945333446583305,
      "ci_metric": "octane_score",
      "median_latency_ms": 3467.8047504276037,
      "p95_latency_ms": 16475.641292519867,
      "input_tokens": 3771,
      "output_tokens": 2805,
      "reasoning_tokens": 14402,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ee1f2730-69f1-4c0f-af48-e66c47a58a21",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 16,
      "content_successes": 16,
      "success_rate": 0.7619047619047619,
      "total_cost_usd": 0.0023356,
      "cost_per_success_usd": 0.0001619576923076923,
      "octane_score": 71.09862499703152,
      "ci_low": 38.2749986791019,
      "ci_high": 116.23326203620283,
      "ci_metric": "octane_score",
      "median_latency_ms": 2784.5872500911355,
      "p95_latency_ms": 7167.283458635211,
      "input_tokens": 2190,
      "output_tokens": 2372,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3e5d5733-c1fc-4786-8bd6-cf08a84c7f00",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.254125595092773,
      "p95_latency_ms": 37.86133322864771,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "ef6b5b71-96f7-462e-97de-01a5c24e7369",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 20,
      "content_successes": 20,
      "success_rate": 0.9523809523809523,
      "total_cost_usd": 0.43868999999999997,
      "cost_per_success_usd": 0.023968615384615383,
      "octane_score": 0.4804186243549257,
      "ci_low": 0.3368660665630628,
      "ci_high": 0.7278947523602746,
      "ci_metric": "octane_score",
      "median_latency_ms": 3193.20279173553,
      "p95_latency_ms": 5679.937917739153,
      "input_tokens": 2934,
      "output_tokens": 4724,
      "reasoning_tokens": 3463,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f744caf5-4e7f-4c8e-ba2c-a4a25ead80d7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 3,
      "content_successes": 3,
      "success_rate": 0.14285714285714285,
      "total_cost_usd": 0.020970040000000002,
      "cost_per_success_usd": 0.012071128333333334,
      "octane_score": 0.9539265023777174,
      "ci_low": 0.6173273944215129,
      "ci_high": 4.796543177533384,
      "ci_metric": "octane_score",
      "median_latency_ms": 19462.106375023723,
      "p95_latency_ms": 40927.12733335793,
      "input_tokens": 2292,
      "output_tokens": 80081,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9fc4751d-edb7-4f87-bb84-7197865c586f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 0.8571428571428571,
      "total_cost_usd": 0.4289928,
      "cost_per_success_usd": 0.028724715789473686,
      "octane_score": 0.40087321716822516,
      "ci_low": 0.12792200641746973,
      "ci_high": 2.1775446253198063,
      "ci_metric": "octane_score",
      "median_latency_ms": 12038.328208960593,
      "p95_latency_ms": 130224.6026257053,
      "input_tokens": 3219,
      "output_tokens": 2178,
      "reasoning_tokens": 68405,
      "undefined_reason": null
    },
    {
      "snapshot_id": "53acb692-9c6b-41db-9ac8-ec4cedb63869",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 0.8571428571428571,
      "total_cost_usd": 0.3141527,
      "cost_per_success_usd": 0.01904291186440678,
      "octane_score": 0.6046853187558952,
      "ci_low": 0.24274381724219832,
      "ci_high": 3.6040357216981422,
      "ci_metric": "octane_score",
      "median_latency_ms": 12438.007917255163,
      "p95_latency_ms": 210270.98483406007,
      "input_tokens": 3134,
      "output_tokens": 25243,
      "reasoning_tokens": 72489,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a4f46c4a-ed26-4659-8cab-2046620b3731",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 18,
      "content_successes": 18,
      "success_rate": 0.8571428571428571,
      "total_cost_usd": 0.42675,
      "cost_per_success_usd": 0.02849042105263158,
      "octane_score": 0.40416985096489566,
      "ci_low": 0.12840002925516453,
      "ci_high": 2.47878409158917,
      "ci_metric": "octane_score",
      "median_latency_ms": 14258.169249631464,
      "p95_latency_ms": 344686.63816712797,
      "input_tokens": 3090,
      "output_tokens": 1961,
      "reasoning_tokens": 68134,
      "undefined_reason": null
    },
    {
      "snapshot_id": "a0809196-e803-45eb-bcb2-b35a883e7617",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 16,
      "content_successes": 16,
      "success_rate": 0.7619047619047619,
      "total_cost_usd": 0.5429955,
      "cost_per_success_usd": 0.03980127,
      "octane_score": 0.28931160314153875,
      "ci_low": 0.145647778963146,
      "ci_high": 0.5117016607249372,
      "ci_metric": "octane_score",
      "median_latency_ms": 148732.7930405736,
      "p95_latency_ms": 230538.04820869118,
      "input_tokens": 2781,
      "output_tokens": 2190,
      "reasoning_tokens": 178345,
      "undefined_reason": null
    },
    {
      "snapshot_id": "92656c70-798c-4033-bbab-11f60762f87f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "coding",
      "task_id": null,
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 21,
      "successes": 17,
      "content_successes": 17,
      "success_rate": 0.8095238095238095,
      "total_cost_usd": 0.393923,
      "cost_per_success_usd": 0.027455415094339623,
      "octane_score": 0.4194061241180516,
      "ci_low": 0.2280968103264855,
      "ci_high": 0.7309835265182854,
      "ci_metric": "octane_score",
      "median_latency_ms": 119543.45616605133,
      "p95_latency_ms": 213743.24529152364,
      "input_tokens": 3096,
      "output_tokens": 2132,
      "reasoning_tokens": 154818,
      "undefined_reason": null
    },
    {
      "snapshot_id": "563fecbe-4ed9-4ef6-a257-4bda1d22c0e2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 6,
      "content_successes": 12,
      "success_rate": 0.4,
      "total_cost_usd": 0.006094,
      "cost_per_success_usd": 0.00148075,
      "octane_score": 3.4482043366054835,
      "ci_low": 0.6262937841646676,
      "ci_high": 16.61795343281413,
      "ci_metric": "octane_score",
      "median_latency_ms": 1022.6468751206994,
      "p95_latency_ms": 1294.0398873761296,
      "input_tokens": 1959,
      "output_tokens": 827,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "bf7965fd-3248-4fc7-a0a4-407fa8c81571",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.042045,
      "cost_per_success_usd": 0.0029635714285714287,
      "octane_score": 1.7228970836346102,
      "ci_low": 0.7509801295012066,
      "ci_high": 2.8894502486574463,
      "ci_metric": "octane_score",
      "median_latency_ms": 1899.5847078040242,
      "p95_latency_ms": 5383.389770798385,
      "input_tokens": 1959,
      "output_tokens": 1290,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3034b677-a658-421c-a097-4d6c1942256c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.021417,
      "cost_per_success_usd": 0.001502785714285714,
      "octane_score": 3.397642473501592,
      "ci_low": 1.8666568848413034,
      "ci_high": 4.933171643370114,
      "ci_metric": "octane_score",
      "median_latency_ms": 1600.10483302176,
      "p95_latency_ms": 4362.1801706030965,
      "input_tokens": 1959,
      "output_tokens": 1036,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "be8b8391-229b-47e0-ace9-98e77c4cc60b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.00051072,
      "cost_per_success_usd": 3.544238095238096e-05,
      "octane_score": 144.06279810288996,
      "ci_low": 55.16699822982524,
      "ci_high": 376.0301665378824,
      "ci_metric": "octane_score",
      "median_latency_ms": 1228.0413340777159,
      "p95_latency_ms": 1730.082750227302,
      "input_tokens": 1812,
      "output_tokens": 148,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ffdbe489-b63e-4303-b89a-27aabb2a718e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.00043852,
      "cost_per_success_usd": 3.067952380952381e-05,
      "octane_score": 166.4278951371319,
      "ci_low": 77.91003013722649,
      "ci_high": 287.8450510538529,
      "ci_metric": "octane_score",
      "median_latency_ms": 2038.7637494131923,
      "p95_latency_ms": 5633.565575256942,
      "input_tokens": 2354,
      "output_tokens": 477,
      "reasoning_tokens": 913,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cd989e99-6b7b-4961-be95-3d465f4c4b6e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.002617453,
      "cost_per_success_usd": 0.00018586169047619047,
      "octane_score": 27.471656791385193,
      "ci_low": 13.807513516706612,
      "ci_high": 44.50960734557493,
      "ci_metric": "octane_score",
      "median_latency_ms": 2793.4467922896147,
      "p95_latency_ms": 5619.979120790958,
      "input_tokens": 2591,
      "output_tokens": 109,
      "reasoning_tokens": 1731,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4aebe975-8087-4063-958e-3cd209d32d14",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.0009044999999999999,
      "cost_per_success_usd": 6.037857142857142e-05,
      "octane_score": 84.56524310895539,
      "ci_low": 33.48468063872256,
      "ci_high": 156.507770519437,
      "ci_metric": "octane_score",
      "median_latency_ms": 645.0532907620072,
      "p95_latency_ms": 824.5804747566578,
      "input_tokens": 1740,
      "output_tokens": 153,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fc01e271-75f9-415d-b7b9-ac67b7a238e7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.046225499999999996,
      "cost_per_success_usd": 0.0031983214285714284,
      "octane_score": 1.5964400969258425,
      "ci_low": 0.6553101473937534,
      "ci_high": 3.331521041913528,
      "ci_metric": "octane_score",
      "median_latency_ms": 2594.638915732503,
      "p95_latency_ms": 3854.485871084034,
      "input_tokens": 1747,
      "output_tokens": 149,
      "reasoning_tokens": 5665,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8f190660-28f2-49dc-a86a-c581fc2bf849",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.0075525,
      "cost_per_success_usd": 0.0005140446428571429,
      "octane_score": 9.932850468101364,
      "ci_low": 3.9133201440718017,
      "ci_high": 18.615929789582268,
      "ci_metric": "octane_score",
      "median_latency_ms": 2505.937167443335,
      "p95_latency_ms": 2948.7505830824366,
      "input_tokens": 1740,
      "output_tokens": 153,
      "reasoning_tokens": 3527,
      "undefined_reason": null
    },
    {
      "snapshot_id": "77487e4a-ed46-4e60-af61-e6086c61bec3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.00297708,
      "cost_per_success_usd": 0.00020449190476190475,
      "octane_score": 24.968854279945877,
      "ci_low": 10.745754736784294,
      "ci_high": 46.128802108205846,
      "ci_metric": "octane_score",
      "median_latency_ms": 7832.034250721335,
      "p95_latency_ms": 50695.66982490938,
      "input_tokens": 1698,
      "output_tokens": 1050,
      "reasoning_tokens": 6186,
      "undefined_reason": null
    },
    {
      "snapshot_id": "27bdd4a9-6e5c-4441-b0bf-ef817171fbea",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 14,
      "content_successes": 14,
      "success_rate": 0.9333333333333333,
      "total_cost_usd": 0.013238759999999999,
      "cost_per_success_usd": 0.000936186,
      "octane_score": 5.453968091200435,
      "ci_low": 1.7328884850764745,
      "ci_high": 13.163009041989039,
      "ci_metric": "octane_score",
      "median_latency_ms": 6714.309125207365,
      "p95_latency_ms": 11707.892208732661,
      "input_tokens": 1719,
      "output_tokens": 217,
      "reasoning_tokens": 6157,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ed96c6ca-d87a-4770-aeb5-54e1e4aa9612",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.0096320964,
      "cost_per_success_usd": 0.0006335126571428572,
      "octane_score": 8.05971043176361,
      "ci_low": 2.7741456730491354,
      "ci_high": 15.739694103468103,
      "ci_metric": "octane_score",
      "median_latency_ms": 2341.8955830857158,
      "p95_latency_ms": 3427.5474664755166,
      "input_tokens": 1803,
      "output_tokens": 230,
      "reasoning_tokens": 2666,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cf2a706b-0381-4e2e-bfa0-0bbd01d63e67",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 13,
      "content_successes": 15,
      "success_rate": 0.8666666666666667,
      "total_cost_usd": 0.01415716,
      "cost_per_success_usd": 0.0011756905882352943,
      "octane_score": 4.342918640773116,
      "ci_low": 1.0048440636287899,
      "ci_high": 12.493596796115364,
      "ci_metric": "octane_score",
      "median_latency_ms": 4135.532332584262,
      "p95_latency_ms": 6424.616829492151,
      "input_tokens": 1803,
      "output_tokens": 400,
      "reasoning_tokens": 2476,
      "undefined_reason": null
    },
    {
      "snapshot_id": "815f292e-dcd0-49da-b135-7326d4d235d8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.0020340000000000002,
      "cost_per_success_usd": 0.0001761818181818182,
      "octane_score": 28.981018723278776,
      "ci_low": 14.627831344317618,
      "ci_high": 38.420101292222945,
      "ci_metric": "octane_score",
      "median_latency_ms": 630.9425411745906,
      "p95_latency_ms": 955.0286464393139,
      "input_tokens": 1692,
      "output_tokens": 170,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "14718f1b-aec0-4f72-a886-f50486ccba91",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.0009408,
      "cost_per_success_usd": 6.368571428571429e-05,
      "octane_score": 80.17384477344099,
      "ci_low": 44.36782786885247,
      "ci_high": 111.1098665734649,
      "ci_metric": "octane_score",
      "median_latency_ms": 888.690416701138,
      "p95_latency_ms": 2666.043312475085,
      "input_tokens": 1692,
      "output_tokens": 159,
      "reasoning_tokens": 343,
      "undefined_reason": null
    },
    {
      "snapshot_id": "97eb4124-ea96-468f-a712-b8b47455bed0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.00792634,
      "cost_per_success_usd": 0.0005368457142857142,
      "octane_score": 9.51097947800911,
      "ci_low": 4.331947500375275,
      "ci_high": 15.127137517487512,
      "ci_metric": "octane_score",
      "median_latency_ms": 2264.756958000362,
      "p95_latency_ms": 4948.425057809798,
      "input_tokens": 30221,
      "output_tokens": 736,
      "reasoning_tokens": 1675,
      "undefined_reason": null
    },
    {
      "snapshot_id": "fd3f4746-71a1-44e4-b7e7-6c22bcac14ba",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.007134,
      "cost_per_success_usd": 0.0004942857142857143,
      "octane_score": 10.329913294797686,
      "ci_low": 5.908549913052801,
      "ci_high": 14.874258656296524,
      "ci_metric": "octane_score",
      "median_latency_ms": 1653.5912919789553,
      "p95_latency_ms": 3500.547629594803,
      "input_tokens": 1692,
      "output_tokens": 153,
      "reasoning_tokens": 222,
      "undefined_reason": null
    },
    {
      "snapshot_id": "49b36ca9-efb4-4a2c-a7e0-1bac4dcdcd34",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.14611549999999998,
      "cost_per_success_usd": 0.009861142857142858,
      "octane_score": 0.51778263892913,
      "ci_low": 0.2322566087471606,
      "ci_high": 0.7784661811052306,
      "ci_metric": "octane_score",
      "median_latency_ms": 2940.087500028312,
      "p95_latency_ms": 6084.853125456721,
      "input_tokens": 29884,
      "output_tokens": 684,
      "reasoning_tokens": 694,
      "undefined_reason": null
    },
    {
      "snapshot_id": "81a0e353-5a56-43c6-8ad7-8dfe452d0b97",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.008076,
      "cost_per_success_usd": 0.0005591428571428571,
      "octane_score": 9.131706693919263,
      "ci_low": 5.542535169779133,
      "ci_high": 12.574873754896167,
      "ci_metric": "octane_score",
      "median_latency_ms": 875.0442089512944,
      "p95_latency_ms": 2638.636082690209,
      "input_tokens": 1692,
      "output_tokens": 154,
      "reasoning_tokens": 237,
      "undefined_reason": null
    },
    {
      "snapshot_id": "08a169d2-260f-4336-8473-c2e72549c7a6",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.030788000000000003,
      "cost_per_success_usd": 0.0021044857142857145,
      "octane_score": 2.426212036873616,
      "ci_low": 1.169986817120378,
      "ci_high": 3.643450068202991,
      "ci_metric": "octane_score",
      "median_latency_ms": 2091.1485832184553,
      "p95_latency_ms": 4288.913584128022,
      "input_tokens": 29915,
      "output_tokens": 715,
      "reasoning_tokens": 1124,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c06402ac-fa2e-48c0-bab6-82c508e0fe1b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.00072019,
      "cost_per_success_usd": 5.1059285714285706e-05,
      "octane_score": 100.0,
      "ci_low": 1.8898319444444446e-05,
      "ci_high": 0.00010747693727106218,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 4485.257084481418,
      "p95_latency_ms": 37169.07335920257,
      "input_tokens": 2592,
      "output_tokens": 268,
      "reasoning_tokens": 3511,
      "undefined_reason": null
    },
    {
      "snapshot_id": "573b6016-2044-4056-9d67-1cb27d8137d7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.024298,
      "cost_per_success_usd": 0.0016701142857142855,
      "octane_score": 3.0572330379443664,
      "ci_low": 1.483690800865946,
      "ci_high": 5.112634615516158,
      "ci_metric": "octane_score",
      "median_latency_ms": 2670.9350841119885,
      "p95_latency_ms": 6223.062733560797,
      "input_tokens": 4719,
      "output_tokens": 104,
      "reasoning_tokens": 3098,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c839c83d-c809-4208-be4b-40cb8f47bad5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.036906,
      "cost_per_success_usd": 0.0025597142857142854,
      "octane_score": 1.9947259738810132,
      "ci_low": 1.0262596417892964,
      "ci_high": 3.3072351617492,
      "ci_metric": "octane_score",
      "median_latency_ms": 3928.6232497543097,
      "p95_latency_ms": 9032.965558301657,
      "input_tokens": 4719,
      "output_tokens": 99,
      "reasoning_tokens": 5247,
      "undefined_reason": null
    },
    {
      "snapshot_id": "dc0cbd40-88ec-4b10-b1fe-3e143c611b7f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.01658078,
      "cost_per_success_usd": 0.001156657380952381,
      "octane_score": 4.414382906738032,
      "ci_low": 2.506452684882769,
      "ci_high": 6.508091050633045,
      "ci_metric": "octane_score",
      "median_latency_ms": 5265.046499669552,
      "p95_latency_ms": 15624.200187437234,
      "input_tokens": 1727,
      "output_tokens": 185,
      "reasoning_tokens": 6978,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6f120ce3-5093-478b-a56f-7d8095097b05",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.0346608,
      "cost_per_success_usd": 0.002336042857142857,
      "octane_score": 2.18571699393969,
      "ci_low": 0.8447897001726415,
      "ci_high": 4.319478127445881,
      "ci_metric": "octane_score",
      "median_latency_ms": 1760.5849578976631,
      "p95_latency_ms": 8242.90379174053,
      "input_tokens": 2936,
      "output_tokens": 356,
      "reasoning_tokens": 1644,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b42df5af-3b25-45fd-b0da-b85e31a5e61e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 11,
      "content_successes": 15,
      "success_rate": 0.7333333333333333,
      "total_cost_usd": 0.0024460000000000003,
      "cost_per_success_usd": 0.0002460689655172414,
      "octane_score": 20.74998999039077,
      "ci_low": 4.295643677221469,
      "ci_high": 54.08097532392633,
      "ci_metric": "octane_score",
      "median_latency_ms": 3174.295208416879,
      "p95_latency_ms": 10163.89669552445,
      "input_tokens": 1758,
      "output_tokens": 2618,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "07b4bb83-8fdb-4de5-b5e1-1fe1b3eaead9",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.174708925187588,
      "p95_latency_ms": 37.84329229965805,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "14a19b96-205e-4773-b03d-498e9fc6fd70",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 15,
      "success_rate": 0.8,
      "total_cost_usd": 0.11395,
      "cost_per_success_usd": 0.011095,
      "octane_score": 0.46020086267945653,
      "ci_low": 0.07214262211676702,
      "ci_high": 1.9846532828446968,
      "ci_metric": "octane_score",
      "median_latency_ms": 2821.426250040531,
      "p95_latency_ms": 3199.7116165235634,
      "input_tokens": 2610,
      "output_tokens": 1014,
      "reasoning_tokens": 743,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0f0cce3d-21c1-40eb-b8e3-31763e399ce4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 9,
      "content_successes": 9,
      "success_rate": 0.6,
      "total_cost_usd": 0.00581906,
      "cost_per_success_usd": 0.0006370674074074075,
      "octane_score": 8.014738333903349,
      "ci_low": 2.3520530688608616,
      "ci_high": 14.541875194556749,
      "ci_metric": "octane_score",
      "median_latency_ms": 4175.9244576096535,
      "p95_latency_ms": 25864.745612256225,
      "input_tokens": 1884,
      "output_tokens": 21910,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "89cd03de-b5c1-43fc-9972-91c8bd5b0328",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.018954,
      "cost_per_success_usd": 0.0013408571428571428,
      "octane_score": 3.8079586618367776,
      "ci_low": 2.0199658629352637,
      "ci_high": 6.049557454617261,
      "ci_metric": "octane_score",
      "median_latency_ms": 2847.6716242730618,
      "p95_latency_ms": 5662.669249530878,
      "input_tokens": 2586,
      "output_tokens": 356,
      "reasoning_tokens": 1941,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c6b38b4d-84eb-417b-b679-dd21a2343351",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.0080269,
      "cost_per_success_usd": 0.0005780190476190476,
      "octane_score": 8.83349535358861,
      "ci_low": 4.42499166560376,
      "ci_high": 14.405308359103117,
      "ci_metric": "octane_score",
      "median_latency_ms": 2843.617457896471,
      "p95_latency_ms": 8486.4454046823,
      "input_tokens": 2506,
      "output_tokens": 664,
      "reasoning_tokens": 1492,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8d631136-fd9a-4fd0-a006-e7ef5e3adc3c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.01968,
      "cost_per_success_usd": 0.0013874285714285714,
      "octane_score": 3.680137973640856,
      "ci_low": 2.234654697956704,
      "ci_high": 5.326949441639437,
      "ci_metric": "octane_score",
      "median_latency_ms": 3518.4792494401336,
      "p95_latency_ms": 7962.499870825559,
      "input_tokens": 2454,
      "output_tokens": 180,
      "reasoning_tokens": 2282,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b96c220e-2064-4c4a-bf43-499680dd9e00",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.0490455,
      "cost_per_success_usd": 0.0035122500000000006,
      "octane_score": 1.4537486145429765,
      "ci_low": 0.9435658795505831,
      "ci_high": 3.691953312034078,
      "ci_metric": "octane_score",
      "median_latency_ms": 3943.950708024204,
      "p95_latency_ms": 64876.623816881314,
      "input_tokens": 2241,
      "output_tokens": 117,
      "reasoning_tokens": 15858,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c16b9527-7b4a-4a90-ba37-dd7e306ab351",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "customer_service",
      "task_id": null,
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.036293,
      "cost_per_success_usd": 0.002580059523809524,
      "octane_score": 1.9789964240396811,
      "ci_low": 1.1153728962393328,
      "ci_high": 6.053912090758758,
      "ci_metric": "octane_score",
      "median_latency_ms": 4609.73387490958,
      "p95_latency_ms": 51905.09571665897,
      "input_tokens": 2466,
      "output_tokens": 117,
      "reasoning_tokens": 13907,
      "undefined_reason": null
    },
    {
      "snapshot_id": "be99a373-bb15-4687-8d33-9f8823e355a2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 6,
      "content_successes": 6,
      "success_rate": 0.4,
      "total_cost_usd": 0.004476,
      "cost_per_success_usd": 0.0007728333333333333,
      "octane_score": 15.22021864269007,
      "ci_low": 5.011521824616595,
      "ci_high": 28.945949225765737,
      "ci_metric": "octane_score",
      "median_latency_ms": 613.8791665434837,
      "p95_latency_ms": 695.8502033725381,
      "input_tokens": 2601,
      "output_tokens": 375,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d42da7a0-259d-4fea-9c74-6237555fa67e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 14,
      "content_successes": 14,
      "success_rate": 0.9333333333333333,
      "total_cost_usd": 0.041979999999999996,
      "cost_per_success_usd": 0.0027925,
      "octane_score": 4.2122443357895465,
      "ci_low": 2.2429694138540106,
      "ci_high": 7.702486360131852,
      "ci_metric": "octane_score",
      "median_latency_ms": 1996.1476251482964,
      "p95_latency_ms": 5831.073937751353,
      "input_tokens": 2601,
      "output_tokens": 1159,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c900b47f-c5f5-4ca8-8167-ca0eff842abf",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 10,
      "content_successes": 10,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.028428000000000002,
      "cost_per_success_usd": 0.0025663714285714284,
      "octane_score": 4.58339435077019,
      "ci_low": 1.4829979835722065,
      "ci_high": 8.924709977853958,
      "ci_metric": "octane_score",
      "median_latency_ms": 2297.8938752785325,
      "p95_latency_ms": 6068.766199983656,
      "input_tokens": 2601,
      "output_tokens": 1375,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b2c9ecc4-cf1c-4d3b-a4dd-6b25bff9e3da",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 6,
      "content_successes": 6,
      "success_rate": 0.4,
      "total_cost_usd": 0.0009368200000000001,
      "cost_per_success_usd": 0.00015100111111111112,
      "octane_score": 77.89805135299281,
      "ci_low": 25.982900596628046,
      "ci_high": 152.50060344618902,
      "ci_metric": "octane_score",
      "median_latency_ms": 1209.402957931161,
      "p95_latency_ms": 10743.705953750748,
      "input_tokens": 2388,
      "output_tokens": 919,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "df0a2631-37cd-4d2e-ae82-15bfe4db5f05",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 9,
      "content_successes": 9,
      "success_rate": 0.6,
      "total_cost_usd": 0.001470616,
      "cost_per_success_usd": 0.00015465163636363635,
      "octane_score": 76.05928126123662,
      "ci_low": 34.64585461747407,
      "ci_high": 136.69215046631012,
      "ci_metric": "octane_score",
      "median_latency_ms": 4550.234083086252,
      "p95_latency_ms": 45314.486642368094,
      "input_tokens": 2735,
      "output_tokens": 1179,
      "reasoning_tokens": 5863,
      "undefined_reason": null
    },
    {
      "snapshot_id": "05b02203-a29f-4fe3-9ae7-99fd201ed8b2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.024803758,
      "cost_per_success_usd": 0.00219560305,
      "octane_score": 5.357385665725098,
      "ci_low": 2.166797492981589,
      "ci_high": 14.178336011992203,
      "ci_metric": "octane_score",
      "median_latency_ms": 13279.925999231637,
      "p95_latency_ms": 92045.18351219583,
      "input_tokens": 3210,
      "output_tokens": 345,
      "reasoning_tokens": 26687,
      "undefined_reason": null
    },
    {
      "snapshot_id": "11552f78-a58b-434e-ac67-e1a8d1409c3c",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 6,
      "content_successes": 6,
      "success_rate": 0.4,
      "total_cost_usd": 0.0016872,
      "cost_per_success_usd": 0.000286,
      "octane_score": 41.128294782140934,
      "ci_low": 13.637675841262778,
      "ci_high": 75.50864627270586,
      "ci_metric": "octane_score",
      "median_latency_ms": 688.2472913712263,
      "p95_latency_ms": 936.5169746801257,
      "input_tokens": 2424,
      "output_tokens": 384,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "774c7007-3362-4e6f-aeab-c7d179617da8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 9,
      "content_successes": 9,
      "success_rate": 0.6,
      "total_cost_usd": 0.081486,
      "cost_per_success_usd": 0.008000272727272729,
      "octane_score": 1.4702864150610246,
      "ci_low": 0.4157003163161245,
      "ci_high": 2.8699985879054184,
      "ci_metric": "octane_score",
      "median_latency_ms": 3678.834166377783,
      "p95_latency_ms": 5452.32839519158,
      "input_tokens": 2434,
      "output_tokens": 378,
      "reasoning_tokens": 10000,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5262e424-9cf4-42a5-b3db-8afd2d58a353",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 11,
      "content_successes": 11,
      "success_rate": 0.7333333333333333,
      "total_cost_usd": 0.014163374999999999,
      "cost_per_success_usd": 0.0012670439189189189,
      "octane_score": 9.283571099673168,
      "ci_low": 3.9483686230855195,
      "ci_high": 19.060747053485123,
      "ci_metric": "octane_score",
      "median_latency_ms": 4254.869583062828,
      "p95_latency_ms": 6693.330679833889,
      "input_tokens": 2424,
      "output_tokens": 373,
      "reasoning_tokens": 6696,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ed53a85f-e6ce-48a2-bbdd-2a5af4877a99",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 4,
      "content_successes": 4,
      "success_rate": 0.26666666666666666,
      "total_cost_usd": 0.0061786,
      "cost_per_success_usd": 0.0014600160000000002,
      "octane_score": 8.05655027595061,
      "ci_low": 2.284250110816777,
      "ci_high": 19.812284128385908,
      "ci_metric": "octane_score",
      "median_latency_ms": 11379.366124980152,
      "p95_latency_ms": 63744.03189523137,
      "input_tokens": 2310,
      "output_tokens": 2072,
      "reasoning_tokens": 13092,
      "undefined_reason": null
    },
    {
      "snapshot_id": "de366306-99ed-4bcb-8d51-11192e8c15de",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 8,
      "content_successes": 8,
      "success_rate": 0.5333333333333333,
      "total_cost_usd": 0.10533576,
      "cost_per_success_usd": 0.014063888888888888,
      "octane_score": 0.8363755146690166,
      "ci_low": 0.14049616924099995,
      "ci_high": 5.906697964339947,
      "ci_metric": "octane_score",
      "median_latency_ms": 18308.998583815992,
      "p95_latency_ms": 153441.9441955164,
      "input_tokens": 2366,
      "output_tokens": 245,
      "reasoning_tokens": 53982,
      "undefined_reason": null
    },
    {
      "snapshot_id": "0e1b1589-8eb8-450e-a0ba-7de26b7dfd50",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.022564794,
      "cost_per_success_usd": 0.0019016598153846153,
      "octane_score": 6.185487126840966,
      "ci_low": 4.275285430130507,
      "ci_high": 9.242266222214452,
      "ci_metric": "octane_score",
      "median_latency_ms": 3602.343915961683,
      "p95_latency_ms": 60425.77619254571,
      "input_tokens": 2415,
      "output_tokens": 595,
      "reasoning_tokens": 6489,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f6232d39-7e7a-456f-8a29-6b826f7302a1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 9,
      "content_successes": 12,
      "success_rate": 0.6,
      "total_cost_usd": 0.02815532,
      "cost_per_success_usd": 0.0035211703703703708,
      "octane_score": 3.340563241890241,
      "ci_low": 0.8072934886322516,
      "ci_high": 7.786944252082864,
      "ci_metric": "octane_score",
      "median_latency_ms": 5683.889999985695,
      "p95_latency_ms": 13849.933428596705,
      "input_tokens": 2415,
      "output_tokens": 510,
      "reasoning_tokens": 5419,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c509ea8a-1c7d-4bb0-a60f-f13b3b6d4231",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 4,
      "content_successes": 4,
      "success_rate": 0.26666666666666666,
      "total_cost_usd": 0.0032535,
      "cost_per_success_usd": 0.0009335454545454546,
      "octane_score": 12.600020974254102,
      "ci_low": 3.2360584818713227,
      "ci_high": 28.91699445922224,
      "ci_metric": "octane_score",
      "median_latency_ms": 699.4461249560118,
      "p95_latency_ms": 913.9299342408775,
      "input_tokens": 2328,
      "output_tokens": 335,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6988012b-1ef7-4f34-87d8-1fac14cef3b2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 13,
      "content_successes": 13,
      "success_rate": 0.8666666666666667,
      "total_cost_usd": 0.0018888,
      "cost_per_success_usd": 0.00015108571428571428,
      "octane_score": 77.85443093215407,
      "ci_low": 41.5001099806723,
      "ci_high": 145.9619566028662,
      "ci_metric": "octane_score",
      "median_latency_ms": 2324.3298335000873,
      "p95_latency_ms": 3160.243804566562,
      "input_tokens": 2328,
      "output_tokens": 306,
      "reasoning_tokens": 880,
      "undefined_reason": null
    },
    {
      "snapshot_id": "40372417-d42a-4d71-844b-9068c8467fa4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 15,
      "content_successes": 15,
      "success_rate": 1.0,
      "total_cost_usd": 0.01327864,
      "cost_per_success_usd": 0.0009131475000000001,
      "octane_score": 12.881481149203504,
      "ci_low": 6.412833304265016,
      "ci_high": 24.211196907233354,
      "ci_metric": "octane_score",
      "median_latency_ms": 3561.8935842067003,
      "p95_latency_ms": 8455.102983210236,
      "input_tokens": 34841,
      "output_tokens": 1216,
      "reasoning_tokens": 4478,
      "undefined_reason": null
    },
    {
      "snapshot_id": "312c4b9f-63c6-48b3-89d9-1d9dc4f5d794",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.013266,
      "cost_per_success_usd": 0.0011217435897435896,
      "octane_score": 10.486079363628052,
      "ci_low": 6.601595674097667,
      "ci_high": 17.45588092847059,
      "ci_metric": "octane_score",
      "median_latency_ms": 1895.2230839058757,
      "p95_latency_ms": 4271.944345533848,
      "input_tokens": 2328,
      "output_tokens": 294,
      "reasoning_tokens": 567,
      "undefined_reason": null
    },
    {
      "snapshot_id": "5707daa2-65de-4038-9ef2-b479145b7d90",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.24299800000000002,
      "cost_per_success_usd": 0.020789641025641025,
      "octane_score": 0.5657958352039231,
      "ci_low": 0.29738941599558966,
      "ci_high": 0.9767226446008709,
      "ci_metric": "octane_score",
      "median_latency_ms": 3795.680374838412,
      "p95_latency_ms": 6643.246487434952,
      "input_tokens": 33905,
      "output_tokens": 1111,
      "reasoning_tokens": 1777,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8fe55619-9436-40f5-af66-9a5bd7cee7e0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.014388,
      "cost_per_success_usd": 0.001224,
      "octane_score": 9.610042735042734,
      "ci_low": 5.118734321965965,
      "ci_high": 15.798451284522562,
      "ci_metric": "octane_score",
      "median_latency_ms": 2345.6469997763634,
      "p95_latency_ms": 2747.216534242034,
      "input_tokens": 2328,
      "output_tokens": 300,
      "reasoning_tokens": 511,
      "undefined_reason": null
    },
    {
      "snapshot_id": "80010bf2-4078-45b5-8b31-f46784b58fb4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.0488244,
      "cost_per_success_usd": 0.004165630769230769,
      "octane_score": 2.8237481810862515,
      "ci_low": 1.354633694931834,
      "ci_high": 4.94383127386123,
      "ci_metric": "octane_score",
      "median_latency_ms": 3179.0420841425657,
      "p95_latency_ms": 4867.489354405552,
      "input_tokens": 33651,
      "output_tokens": 1155,
      "reasoning_tokens": 2273,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b460241b-0b85-4b50-baed-54bb621d6154",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.00139495,
      "cost_per_success_usd": 0.00011762692307692307,
      "octane_score": 100.0,
      "ci_low": 4.7644624999999996e-05,
      "ci_high": 0.00023750134090909058,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 5186.99570838362,
      "p95_latency_ms": 20395.757512282576,
      "input_tokens": 3154,
      "output_tokens": 856,
      "reasoning_tokens": 6793,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e589f1ce-74a7-46f4-accc-d41b196dde6e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 14,
      "content_successes": 14,
      "success_rate": 0.9333333333333333,
      "total_cost_usd": 0.0468772,
      "cost_per_success_usd": 0.0034652355555555557,
      "octane_score": 3.394485632826332,
      "ci_low": 2.074650821504339,
      "ci_high": 5.6066691351758,
      "ci_metric": "octane_score",
      "median_latency_ms": 7492.977083660662,
      "p95_latency_ms": 14185.981345828615,
      "input_tokens": 5319,
      "output_tokens": 258,
      "reasoning_tokens": 6616,
      "undefined_reason": null
    },
    {
      "snapshot_id": "9eca79ba-2bb9-4f14-9b80-8c7038b72277",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 13,
      "content_successes": 13,
      "success_rate": 0.8666666666666667,
      "total_cost_usd": 0.08655,
      "cost_per_success_usd": 0.006888,
      "octane_score": 1.70770794246404,
      "ci_low": 1.0314596310246724,
      "ci_high": 3.0173038733698743,
      "ci_metric": "octane_score",
      "median_latency_ms": 12190.569249913096,
      "p95_latency_ms": 25987.40312904119,
      "input_tokens": 5319,
      "output_tokens": 225,
      "reasoning_tokens": 13195,
      "undefined_reason": null
    },
    {
      "snapshot_id": "21ade2a8-2b96-4ac1-bc69-848e92d8b599",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.05624517,
      "cost_per_success_usd": 0.004942556923076923,
      "octane_score": 2.379879987374956,
      "ci_low": 1.3631302663021412,
      "ci_high": 4.343314741697166,
      "ci_metric": "octane_score",
      "median_latency_ms": 20596.06095775962,
      "p95_latency_ms": 56336.64112500846,
      "input_tokens": 2347,
      "output_tokens": 1494,
      "reasoning_tokens": 23266,
      "undefined_reason": null
    },
    {
      "snapshot_id": "3befcb63-028c-4c07-a20e-7190fc382324",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 13,
      "content_successes": 15,
      "success_rate": 0.8666666666666667,
      "total_cost_usd": 0.1050546,
      "cost_per_success_usd": 0.00884442,
      "octane_score": 1.3299563236133412,
      "ci_low": 0.5156944328868991,
      "ci_high": 3.6725260247123286,
      "ci_metric": "octane_score",
      "median_latency_ms": 2561.403624713421,
      "p95_latency_ms": 5698.4551462344825,
      "input_tokens": 3389,
      "output_tokens": 1532,
      "reasoning_tokens": 5174,
      "undefined_reason": null
    },
    {
      "snapshot_id": "22f39619-466c-47da-85ed-c704029f1c44",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 7,
      "content_successes": 8,
      "success_rate": 0.4666666666666667,
      "total_cost_usd": 0.002771,
      "cost_per_success_usd": 0.0004905900000000001,
      "octane_score": 23.97662469208974,
      "ci_low": 5.535235438638589,
      "ci_high": 72.32201593153923,
      "ci_metric": "octane_score",
      "median_latency_ms": 2778.3513348549604,
      "p95_latency_ms": 15966.31032498553,
      "input_tokens": 2403,
      "output_tokens": 2863,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e62fb7a7-6c89-4b4a-987d-bc5caf3f8287",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.41104134172201,
      "p95_latency_ms": 36.95990806445479,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "f8440b01-cb97-45c7-810c-5acef0223e8e",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 11,
      "content_successes": 15,
      "success_rate": 0.7333333333333333,
      "total_cost_usd": 0.21767999999999998,
      "cost_per_success_usd": 0.0230684375,
      "octane_score": 0.5099041626764842,
      "ci_low": 0.13747017941855913,
      "ci_high": 1.5883469978115472,
      "ci_metric": "octane_score",
      "median_latency_ms": 3163.3627079427242,
      "p95_latency_ms": 5221.766091696918,
      "input_tokens": 3648,
      "output_tokens": 1265,
      "reasoning_tokens": 2359,
      "undefined_reason": null
    },
    {
      "snapshot_id": "295abb8a-1490-4c76-8c68-a137297efd8f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 8,
      "content_successes": 8,
      "success_rate": 0.5333333333333333,
      "total_cost_usd": 0.01266213,
      "cost_per_success_usd": 0.0015506191071428571,
      "octane_score": 7.585803795082876,
      "ci_low": 3.2754742638217818,
      "ci_high": 14.81418558859151,
      "ci_metric": "octane_score",
      "median_latency_ms": 10213.242458179593,
      "p95_latency_ms": 32876.02330353111,
      "input_tokens": 2574,
      "output_tokens": 48057,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c1fc9bbb-eaba-4e00-aa77-31e914fa215f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 14,
      "content_successes": 14,
      "success_rate": 0.9333333333333333,
      "total_cost_usd": 0.090084,
      "cost_per_success_usd": 0.006920999999999999,
      "octane_score": 1.6995654251831107,
      "ci_low": 0.6806141767030581,
      "ci_high": 5.220796334564366,
      "ci_metric": "octane_score",
      "median_latency_ms": 6740.600375458598,
      "p95_latency_ms": 44948.643166478614,
      "input_tokens": 3276,
      "output_tokens": 242,
      "reasoning_tokens": 13680,
      "undefined_reason": null
    },
    {
      "snapshot_id": "327537bd-6998-4ad9-854f-e22a949bfb8a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.0296832,
      "cost_per_success_usd": 0.002650819230769231,
      "octane_score": 4.4373800262038,
      "ci_low": 2.494007034107225,
      "ci_high": 10.758752614392307,
      "ci_metric": "octane_score",
      "median_latency_ms": 8190.494832582772,
      "p95_latency_ms": 20465.97902448848,
      "input_tokens": 3200,
      "output_tokens": 1526,
      "reasoning_tokens": 7300,
      "undefined_reason": null
    },
    {
      "snapshot_id": "e9800306-b19a-41b8-88a4-e20144ac2996",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 12,
      "content_successes": 12,
      "success_rate": 0.8,
      "total_cost_usd": 0.164748,
      "cost_per_success_usd": 0.015433076923076924,
      "octane_score": 0.7621741514230174,
      "ci_low": 0.3208881977007673,
      "ci_high": 5.275785749666341,
      "ci_metric": "octane_score",
      "median_latency_ms": 9149.834458716214,
      "p95_latency_ms": 100989.45793425657,
      "input_tokens": 3144,
      "output_tokens": 386,
      "reasoning_tokens": 26024,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ca560853-11b3-4d64-837c-8b557051cc4b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 10,
      "content_successes": 10,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.2039055,
      "cost_per_success_usd": 0.021434117647058826,
      "octane_score": 0.5487836029166507,
      "ci_low": 0.19575267182381068,
      "ci_high": 1.5731307560484948,
      "ci_metric": "octane_score",
      "median_latency_ms": 53632.12858326733,
      "p95_latency_ms": 147236.5912709384,
      "input_tokens": 2943,
      "output_tokens": 536,
      "reasoning_tokens": 66942,
      "undefined_reason": null
    },
    {
      "snapshot_id": "454201d4-d5b9-4845-ac6d-d24f897c5637",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "legal",
      "task_id": null,
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 15,
      "successes": 8,
      "content_successes": 8,
      "success_rate": 0.5333333333333333,
      "total_cost_usd": 0.20445649999999999,
      "cost_per_success_usd": 0.025367120689655173,
      "octane_score": 0.4636983618124696,
      "ci_low": 0.13331121120553535,
      "ci_high": 1.6816435880022704,
      "ci_metric": "octane_score",
      "median_latency_ms": 53060.497250407934,
      "p95_latency_ms": 209191.16362975904,
      "input_tokens": 3168,
      "output_tokens": 288,
      "reasoning_tokens": 80861,
      "undefined_reason": null
    },
    {
      "snapshot_id": "61b43d53-d427-4616-8f2a-ea024389d73f",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "anthropic:haiku-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 24,
      "content_successes": 24,
      "success_rate": 0.8888888888888888,
      "total_cost_usd": 0.009246,
      "cost_per_success_usd": 0.00042569230769230764,
      "octane_score": 6.9756695980403975,
      "ci_low": 3.886848475769896,
      "ci_high": 15.942419256819491,
      "ci_metric": "octane_score",
      "median_latency_ms": 673.770958557725,
      "p95_latency_ms": 2203.6862458102405,
      "input_tokens": 3111,
      "output_tokens": 1227,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ee30c96d-03d5-474f-9c2f-2dfb11909d18",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "anthropic:opus-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.050505,
      "cost_per_success_usd": 0.0019527222222222224,
      "octane_score": 1.5206919115764317,
      "ci_low": 0.8385217457892594,
      "ci_high": 3.7058993618155527,
      "ci_metric": "octane_score",
      "median_latency_ms": 1785.3194177150726,
      "p95_latency_ms": 4639.024887513369,
      "input_tokens": 3111,
      "output_tokens": 1398,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "08601a3b-9a25-4ca1-920f-cfbd2051e86b",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "anthropic:sonnet-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-09",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.027978,
      "cost_per_success_usd": 0.0011174666666666667,
      "octane_score": 2.6573400151135504,
      "ci_low": 1.7927687665487544,
      "ci_high": 4.344714137569999,
      "ci_metric": "octane_score",
      "median_latency_ms": 1758.6637074127793,
      "p95_latency_ms": 3660.513558331877,
      "input_tokens": 3111,
      "output_tokens": 1243,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4a5676b3-7b5e-4408-93c2-32a351ff1eba",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:deepseek-v3.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 26,
      "content_successes": 26,
      "success_rate": 0.9629629629629629,
      "total_cost_usd": 0.00098196,
      "cost_per_success_usd": 4.0127209302325584e-05,
      "octane_score": 74.00187903714479,
      "ci_low": 52.168542387176004,
      "ci_high": 111.76622686585559,
      "ci_metric": "octane_score",
      "median_latency_ms": 1248.6495422199368,
      "p95_latency_ms": 3299.217820633203,
      "input_tokens": 2736,
      "output_tokens": 734,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "c15786b8-8c1e-45d6-8ae1-9991cc5501d8",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:deepseek-v4-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.00071786,
      "cost_per_success_usd": 2.7503777777777778e-05,
      "octane_score": 107.96658236848272,
      "ci_low": 85.58408829452792,
      "ci_high": 136.49706898723795,
      "ci_metric": "octane_score",
      "median_latency_ms": 2204.0067920461297,
      "p95_latency_ms": 7091.598145850003,
      "input_tokens": 3364,
      "output_tokens": 826,
      "reasoning_tokens": 1667,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2b3e5d54-3d6f-43c2-bc41-971cad0239ff",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:deepseek-v4-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 26,
      "content_successes": 26,
      "success_rate": 0.9629629629629629,
      "total_cost_usd": 0.00379146,
      "cost_per_success_usd": 0.0001539495348837209,
      "octane_score": 19.288716209060087,
      "ci_low": 13.696092292210649,
      "ci_high": 26.753937157249723,
      "ci_metric": "octane_score",
      "median_latency_ms": 2307.140249758959,
      "p95_latency_ms": 4428.913949802518,
      "input_tokens": 3996,
      "output_tokens": 549,
      "reasoning_tokens": 1811,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4fc94b52-71ae-4957-8a27-b91b0ed93c74",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gemini-2.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 25,
      "content_successes": 25,
      "success_rate": 0.9259259259259259,
      "total_cost_usd": 0.0030908,
      "cost_per_success_usd": 0.00013340365853658538,
      "octane_score": 22.25942617664057,
      "ci_low": 11.051973770268205,
      "ci_high": 61.799829921313254,
      "ci_metric": "octane_score",
      "median_latency_ms": 653.757625259459,
      "p95_latency_ms": 1365.7857994548976,
      "input_tokens": 2661,
      "output_tokens": 917,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "214689a3-f295-48f5-be55-f59470ea4479",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gemini-3.6-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.081273,
      "cost_per_success_usd": 0.003126583333333333,
      "octane_score": 0.9497552351253143,
      "ci_low": 0.7098940508524761,
      "ci_high": 1.3154449413702105,
      "ci_metric": "octane_score",
      "median_latency_ms": 2700.751166790724,
      "p95_latency_ms": 3824.9047092162073,
      "input_tokens": 2672,
      "output_tokens": 270,
      "reasoning_tokens": 10032,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d476cf52-9b1e-4b83-9f3e-fdf7169328a2",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gemini-3.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.01343475,
      "cost_per_success_usd": 0.0005065708333333332,
      "octane_score": 5.861942088827109,
      "ci_low": 4.4647981607398695,
      "ci_high": 7.431131270453383,
      "ci_metric": "octane_score",
      "median_latency_ms": 2619.0644996240735,
      "p95_latency_ms": 3383.474679570645,
      "input_tokens": 2661,
      "output_tokens": 285,
      "reasoning_tokens": 6348,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b11649fd-f198-422e-91a3-cf1a95502e61",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:glm-4.7-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 25,
      "content_successes": 25,
      "success_rate": 0.9259259259259259,
      "total_cost_usd": 0.01240718,
      "cost_per_success_usd": 0.0003909655813953489,
      "octane_score": 7.595269328545081,
      "ci_low": 2.658739082621743,
      "ci_high": 15.213545409483782,
      "ci_metric": "octane_score",
      "median_latency_ms": 7632.486958056688,
      "p95_latency_ms": 39689.44245046003,
      "input_tokens": 2613,
      "output_tokens": 4888,
      "reasoning_tokens": 25826,
      "undefined_reason": null
    },
    {
      "snapshot_id": "f6447b6b-0634-4846-87fd-1e0b0fc88f42",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:glm-5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.03027096,
      "cost_per_success_usd": 0.0011376039999999997,
      "octane_score": 2.610301026445837,
      "ci_low": 1.9324435866915723,
      "ci_high": 3.9250563348627727,
      "ci_metric": "octane_score",
      "median_latency_ms": 7899.44841619581,
      "p95_latency_ms": 69304.13797935471,
      "input_tokens": 2662,
      "output_tokens": 780,
      "reasoning_tokens": 14200,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b2ff9a35-4b3c-4f7d-a3f9-8f5654c07396",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:glm-5.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.0186339192,
      "cost_per_success_usd": 0.0006782798133333334,
      "octane_score": 4.377970316255845,
      "ci_low": 3.034404464580923,
      "ci_high": 6.740714746849015,
      "ci_metric": "octane_score",
      "median_latency_ms": 2389.4594591110945,
      "p95_latency_ms": 5316.060137748717,
      "input_tokens": 2802,
      "output_tokens": 352,
      "reasoning_tokens": 5406,
      "undefined_reason": null
    },
    {
      "snapshot_id": "da77e42a-4219-4ee3-8168-7c16d751f414",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:glm-5.3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 24,
      "content_successes": 27,
      "success_rate": 0.8888888888888888,
      "total_cost_usd": 0.01604272,
      "cost_per_success_usd": 0.0006893056410256412,
      "octane_score": 4.307942242385375,
      "ci_low": 2.385117415024857,
      "ci_high": 6.668816504599408,
      "ci_metric": "octane_score",
      "median_latency_ms": 2472.1055841073394,
      "p95_latency_ms": 4555.613949708641,
      "input_tokens": 2802,
      "output_tokens": 477,
      "reasoning_tokens": 2576,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b1d3ee6d-93db-4b77-b806-6a3e5d253cf3",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gpt-5.4-mini",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 24,
      "content_successes": 24,
      "success_rate": 0.8888888888888888,
      "total_cost_usd": 0.00349875,
      "cost_per_success_usd": 0.00014965384615384618,
      "octane_score": 19.84238270653074,
      "ci_low": 12.88317633745675,
      "ci_high": 26.78144795715425,
      "ci_metric": "octane_score",
      "median_latency_ms": 629.5194160193205,
      "p95_latency_ms": 938.4615670889616,
      "input_tokens": 2619,
      "output_tokens": 341,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "695f96e0-bd00-40b7-8deb-e63a1e72e862",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.0019506,
      "cost_per_success_usd": 7.566666666666667e-05,
      "octane_score": 39.244346549192365,
      "ci_low": 30.670601356659606,
      "ci_high": 50.02323392829036,
      "ci_metric": "octane_score",
      "median_latency_ms": 1850.127000361681,
      "p95_latency_ms": 2411.7442751303315,
      "input_tokens": 2619,
      "output_tokens": 316,
      "reasoning_tokens": 873,
      "undefined_reason": null
    },
    {
      "snapshot_id": "b403c93a-fe60-42fe-bd1a-040a17ed8bb4",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gpt-5.6-luna-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.015137920000000001,
      "cost_per_success_usd": 0.0005838573333333334,
      "octane_score": 5.08598371443861,
      "ci_low": 3.74170933322635,
      "ci_high": 6.762884532867001,
      "ci_metric": "octane_score",
      "median_latency_ms": 3181.932542473078,
      "p95_latency_ms": 3827.1550874225795,
      "input_tokens": 53144,
      "output_tokens": 1393,
      "reasoning_tokens": 3174,
      "undefined_reason": null
    },
    {
      "snapshot_id": "88b5f707-8d10-4ac4-b90c-a7786c77bcc5",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-22",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.012317999999999999,
      "cost_per_success_usd": 0.00048344444444444444,
      "octane_score": 6.14235807860262,
      "ci_low": 5.023717207174795,
      "ci_high": 7.551472519396508,
      "ci_metric": "octane_score",
      "median_latency_ms": 1482.2908751666546,
      "p95_latency_ms": 4474.462975654751,
      "input_tokens": 2619,
      "output_tokens": 288,
      "reasoning_tokens": 420,
      "undefined_reason": null
    },
    {
      "snapshot_id": "58034242-5e2b-4cb0-84dd-9b45a1052fae",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gpt-5.6-sol-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.2547435,
      "cost_per_success_usd": 0.009890105555555554,
      "octane_score": 0.30024845257802557,
      "ci_low": 0.24275950210978264,
      "ci_high": 0.3701971500346463,
      "ci_metric": "octane_score",
      "median_latency_ms": 3076.077874749899,
      "p95_latency_ms": 5496.266003325582,
      "input_tokens": 51843,
      "output_tokens": 1259,
      "reasoning_tokens": 1089,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ce946888-a9b8-4bd9-a79d-0c1106efee05",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-18",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.015966,
      "cost_per_success_usd": 0.0006285333333333333,
      "octane_score": 4.724473200395984,
      "ci_low": 3.6312267962807403,
      "ci_high": 6.103011495136981,
      "ci_metric": "octane_score",
      "median_latency_ms": 1937.0015831664205,
      "p95_latency_ms": 2824.389733374119,
      "input_tokens": 2619,
      "output_tokens": 302,
      "reasoning_tokens": 592,
      "undefined_reason": null
    },
    {
      "snapshot_id": "39a3f2dd-4642-41cf-bc89-cab3e47797e7",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gpt-5.6-terra-pro",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.0674614,
      "cost_per_success_usd": 0.0026419377777777775,
      "octane_score": 1.1239813873991482,
      "ci_low": 0.8296573949417816,
      "ci_high": 1.508362809672424,
      "ci_metric": "octane_score",
      "median_latency_ms": 2298.417334444821,
      "p95_latency_ms": 4001.93337071687,
      "input_tokens": 52387,
      "output_tokens": 1361,
      "reasoning_tokens": 2391,
      "undefined_reason": null
    },
    {
      "snapshot_id": "d563b4bc-50c3-46c7-92ea-6011f2609b8a",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:gpt-oss-120b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.00076397,
      "cost_per_success_usd": 2.969488888888889e-05,
      "octane_score": 100.0,
      "ci_low": 2.0193913359788363e-05,
      "ci_high": 4.2258228720238076e-05,
      "ci_metric": "cost_per_success_usd",
      "median_latency_ms": 2571.419416926801,
      "p95_latency_ms": 9371.134741697459,
      "input_tokens": 4142,
      "output_tokens": 1252,
      "reasoning_tokens": 2511,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ac182f06-d07d-4374-ac4d-394cc3309a12",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:grok-4.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.0441144,
      "cost_per_success_usd": 0.0017202666666666666,
      "octane_score": 1.7261794037100193,
      "ci_low": 1.3893575900082884,
      "ci_high": 2.1292123444061644,
      "ci_metric": "octane_score",
      "median_latency_ms": 3349.8846674337983,
      "p95_latency_ms": 6219.912542030215,
      "input_tokens": 8058,
      "output_tokens": 210,
      "reasoning_tokens": 5762,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6c78e1d5-952c-4f17-ac8b-0192c4ff0bb1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:grok-4.6",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.052110000000000004,
      "cost_per_success_usd": 0.0020115333333333334,
      "octane_score": 1.4762315094068648,
      "ci_low": 1.2242360743126373,
      "ci_high": 1.7588205063923283,
      "ci_metric": "octane_score",
      "median_latency_ms": 3435.172749683261,
      "p95_latency_ms": 6930.007086973636,
      "input_tokens": 8058,
      "output_tokens": 195,
      "reasoning_tokens": 7052,
      "undefined_reason": null
    },
    {
      "snapshot_id": "953ac2f1-f02e-4644-a4e3-5fa75ae0b18d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:kimi-k2.5",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-19",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.03225122,
      "cost_per_success_usd": 0.0013168359999999998,
      "octane_score": 2.255018004435548,
      "ci_low": 1.262502522004368,
      "ci_high": 4.185798792776349,
      "ci_metric": "octane_score",
      "median_latency_ms": 6409.872000105679,
      "p95_latency_ms": 14680.110407806931,
      "input_tokens": 2729,
      "output_tokens": 1312,
      "reasoning_tokens": 12597,
      "undefined_reason": null
    },
    {
      "snapshot_id": "dbc9cdfe-f222-4e18-badb-0b278dc9289d",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:kimi-k3",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.055842,
      "cost_per_success_usd": 0.0020348066666666664,
      "octane_score": 1.4593469431439297,
      "ci_low": 1.0997264853228008,
      "ci_high": 1.7805278114499923,
      "ci_metric": "octane_score",
      "median_latency_ms": 1261.6532500833273,
      "p95_latency_ms": 3264.7595662623644,
      "input_tokens": 4661,
      "output_tokens": 1295,
      "reasoning_tokens": 2014,
      "undefined_reason": null
    },
    {
      "snapshot_id": "4f02112c-ed52-444f-b599-3beb5a4a85b0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:llama-4-maverick",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-15",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 13,
      "content_successes": 27,
      "success_rate": 0.48148148148148145,
      "total_cost_usd": 0.0033558,
      "cost_per_success_usd": 0.0003438648648648649,
      "octane_score": 8.635627516221716,
      "ci_low": 2.1605963187820283,
      "ci_high": 26.72016623198087,
      "ci_metric": "octane_score",
      "median_latency_ms": 894.8392914608121,
      "p95_latency_ms": 4359.521878976375,
      "input_tokens": 2715,
      "output_tokens": 3516,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8b8a4381-e47d-450a-a3c8-8f52d27ea0b0",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:muse-spark-1.2",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 0,
      "content_successes": 0,
      "success_rate": 0.0,
      "total_cost_usd": 0.0,
      "cost_per_success_usd": null,
      "octane_score": null,
      "ci_low": null,
      "ci_high": null,
      "ci_metric": null,
      "median_latency_ms": 30.043333768844604,
      "p95_latency_ms": 33.870395924896,
      "input_tokens": 0,
      "output_tokens": 0,
      "reasoning_tokens": 0,
      "undefined_reason": "zero_successes"
    },
    {
      "snapshot_id": "1573ba7a-fc3d-4808-b559-f475a69131be",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:opus-5-fast",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 18,
      "content_successes": 27,
      "success_rate": 0.6666666666666666,
      "total_cost_usd": 0.16246,
      "cost_per_success_usd": 0.010092631578947367,
      "octane_score": 0.2942234506095583,
      "ci_low": 0.09055339946804897,
      "ci_high": 0.6798286019213216,
      "ci_metric": "octane_score",
      "median_latency_ms": 2837.789417244494,
      "p95_latency_ms": 3742.053249757737,
      "input_tokens": 4101,
      "output_tokens": 1182,
      "reasoning_tokens": 1247,
      "undefined_reason": null
    },
    {
      "snapshot_id": "cbe85847-26aa-4469-a03e-1f2881b998c1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:qwen3.5-flash",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-07-26",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 21,
      "content_successes": 21,
      "success_rate": 0.7777777777777778,
      "total_cost_usd": 0.008913645000000001,
      "cost_per_success_usd": 0.00043798820000000006,
      "octane_score": 6.779837650623667,
      "ci_low": 2.942094472044887,
      "ci_high": 16.97164611271945,
      "ci_metric": "octane_score",
      "median_latency_ms": 4023.5884999856353,
      "p95_latency_ms": 12032.348467130209,
      "input_tokens": 2973,
      "output_tokens": 33540,
      "reasoning_tokens": 0,
      "undefined_reason": null
    },
    {
      "snapshot_id": "ddec05c6-bd2e-4197-9bee-cde6896101a1",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:qwen3.8-2.4t",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-12",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.02979,
      "cost_per_success_usd": 0.0011550666666666667,
      "octane_score": 2.5708376620878064,
      "ci_low": 2.023970082500818,
      "ci_high": 3.1852809105314384,
      "ci_metric": "octane_score",
      "median_latency_ms": 2999.853542074561,
      "p95_latency_ms": 4598.801617324352,
      "input_tokens": 4233,
      "output_tokens": 659,
      "reasoning_tokens": 2895,
      "undefined_reason": null
    },
    {
      "snapshot_id": "6744d540-a951-434c-86f7-3bcba7a46a63",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:qwen3.8-27b",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.013533350000000001,
      "cost_per_success_usd": 0.0005119816666666668,
      "octane_score": 5.7999906680686255,
      "ci_low": 4.775923918415329,
      "ci_high": 6.85817405227238,
      "ci_metric": "octane_score",
      "median_latency_ms": 3310.5639591813087,
      "p95_latency_ms": 10363.8292953372,
      "input_tokens": 4083,
      "output_tokens": 1780,
      "reasoning_tokens": 1875,
      "undefined_reason": null
    },
    {
      "snapshot_id": "2ee2e817-bf25-425d-9913-45afbe7c8506",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:qwen3.8-max",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-01",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.028476,
      "cost_per_success_usd": 0.0010861333333333331,
      "octane_score": 2.7340003273590314,
      "ci_low": 2.093449312225275,
      "ci_high": 3.436556415733163,
      "ci_metric": "octane_score",
      "median_latency_ms": 3662.788917310536,
      "p95_latency_ms": 5926.2066752649835,
      "input_tokens": 3999,
      "output_tokens": 383,
      "reasoning_tokens": 3030,
      "undefined_reason": null
    },
    {
      "snapshot_id": "891d3282-d521-4c45-8588-eb677875d184",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:seed-2.0-code",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 26,
      "content_successes": 27,
      "success_rate": 0.9629629629629629,
      "total_cost_usd": 0.065811,
      "cost_per_success_usd": 0.0028528779069767443,
      "octane_score": 1.040874858901943,
      "ci_low": 0.5070484186639603,
      "ci_high": 2.7741861218343837,
      "ci_metric": "octane_score",
      "median_latency_ms": 7539.961125701666,
      "p95_latency_ms": 49467.58208749813,
      "input_tokens": 3540,
      "output_tokens": 558,
      "reasoning_tokens": 20789,
      "undefined_reason": null
    },
    {
      "snapshot_id": "8b3f93f5-9fef-4be9-86a9-a2b9bdf71410",
      "benchmark_version": "0.3.0",
      "calculated_at": "2026-08-22T15:48:07.400963",
      "experiment_id": "weekly-board",
      "scope_level": "vertical",
      "vertical": "medical",
      "task_id": null,
      "model_key": "or:seed-2.1-turbo",
      "baseline_model_key": "or:gpt-oss-120b",
      "pricing_version": "2026-08-21",
      "cost_basis": "billed",
      "task_basket": [
        "banking-day-count-convention@1",
        "banking-effective-annual-rate@1",
        "banking-fx-missing-rate-refusal@1",
        "banking-kyc-risk-tier@1",
        "banking-payment-waterfall@1",
        "banking-statement-field-extraction@1",
        "coding-add-months@1",
        "coding-dedup-implementation@1",
        "coding-dedup-preserve-order@1",
        "coding-merge-intervals@1",
        "coding-pair-sums-performance@1",
        "coding-parse-duration-contract@1",
        "coding-subtract-intervals@1",
        "cs-defective-return-exception@1",
        "cs-escalation-decision@1",
        "cs-prorated-refund@1",
        "cs-refund-eligibility@1",
        "cs-ticket-intent-classification@1",
        "legal-clause-precedence@1",
        "legal-governing-law-extraction@1",
        "legal-liability-cap-carveout@1",
        "legal-notice-period-calculation@1",
        "legal-termination-earliest-date@1",
        "medical-conditional-interaction@1",
        "medical-icd10-lookup@2",
        "medical-infusion-rate-units@1",
        "medical-interaction-extraction@1",
        "medical-max-dose-cap@1",
        "medical-missing-weight-refusal@1",
        "medical-paediatric-dose-volume@1",
        "medical-reference-range-interpretation@1",
        "medical-renal-dose-adjustment@1"
      ],
      "attempts": 27,
      "successes": 27,
      "content_successes": 27,
      "success_rate": 1.0,
      "total_cost_usd": 0.04995,
      "cost_per_success_usd": 0.002089077777777778,
      "octane_score": 1.4214352957445338,
      "ci_low": 0.7661326671078879,
      "ci_high": 3.4488953637140973,
      "ci_metric": "octane_score",
      "median_latency_ms": 6279.076457954943,
      "p95_latency_ms": 36239.11904189736,
      "input_tokens": 3945,
      "output_tokens": 222,
      "reasoning_tokens": 18969,
      "undefined_reason": null
    }
  ]
}