{
 "note": "Receipts copied from public-runs/routing. Every attempt is kept, failures included.",
 "batches": [
  {
   "batch": "batch-1",
   "schema": "agent-public-bench/routing@1",
   "route": "routing",
   "startedAt": null,
   "finishedAt": "2026-10-05T21:07:24.072Z",
   "stopReason": null,
   "trimmed": [],
   "generatedAt": "2026-10-05T21:07:24.071Z",
   "experimentA": {
    "title": "System One decision quality: Jev vs Claude models as routers",
    "caseSets": [
     {
      "purpose": "control.failure_class",
      "label": "Failure class",
      "version": "failure-class-2026-10-05.1",
      "askedCases": 18
     },
     {
      "purpose": "frontdoor.intent",
      "label": "Message intent",
      "version": "frontdoor-intent-2026-10-04.2",
      "askedCases": 20
     },
     {
      "purpose": "memory.is_rule",
      "label": "Is it a rule?",
      "version": "memory-is-rule-2026-10-04.2",
      "askedCases": 12
     },
     {
      "purpose": "context.shape",
      "label": "Context shape",
      "version": "context-shape-2026-10-05.1",
      "askedCases": 32
     }
    ],
    "metric": "Exact: every question the router was asked in the case got an acceptable answer (the production scope of the runner; keys the rule answers are not scored). Key accuracy: the questions the router itself was asked; an unanswered key counts as wrong. Wilson 95% intervals.",
    "caveats": [
     "The case sets and question wording were revised in fix waves against Jev answers on 2026-10-04 and 2026-10-05, so Jev has a home advantage.",
     "Jev numbers are from a recorded production run with the same case versions and the same runner; they were not re-measured here.",
     "Claude routers ran through the Claude Code CLI; the CLI adds startup time and tool-schema tokens a direct API call would not. Haiku 4.5 ran with the CLI default extended thinking; Sonnet 5.5 at effort low, as production asks.",
     "One sample per decision; production asks a second sample when confidence is low. Confidence-gated coverage is therefore not compared.",
     "82 cases in four small hand-labelled sets: intervals are wide."
    ]
   },
   "experimentB": {
    "title": "Routing economics (thought experiment)",
    "kind": "thought-experiment",
    "assumptions": [
     "Same tokens, same cache-read share and same number of calls on every model; a different model would take a different path and number of turns.",
     "Cache writes are not recorded; uncached input is priced at the plain input rate (cache writes cost 1.25x), so every figure is a slight underestimate.",
     "Tokenizers differ between model families; token counts are not converted.",
     "runtime.compact calls record a cost but no tokens (105 calls); they are left out of the repricing.",
     "Policy mix: tiers from the platform routing policy (act, research, verify, review on the strong tier; learning, memory and compact on economy; other purposes standard), with the claude-code ladder economy=Haiku, standard=Sonnet, strong=Opus. The ladder alias \"opus\" is priced as Opus 5.5 here.",
     "Haiku 4.5 has a 200K window; 2 recorded calls had more input than that and could not run on it as recorded."
    ]
   }
  }
 ],
 "receipts": [
  {
   "batch": "batch-1",
   "type": "router",
   "id": "jev",
   "label": "Jev 1.13 (TypeSafe)",
   "kind": "decision model",
   "measured": "recorded production run, 2026-10-05",
   "calls": 82,
   "tokensPerDecision": {
    "input": 802.8,
    "output": 0
   },
   "listCostPer1000Usd": 0.0337,
   "costSource": "provider-reported",
   "latencyMs": null,
   "latencyNote": "Not measured in this benchmark: the recorded report has no per-call latency and no Jev key is available on the bench machine.",
   "pooled": {
    "cases": 82,
    "exact": 74,
    "exactRate": 0.9024,
    "exactCi": [
     0.8191,
     0.9497
    ],
    "keys": 194,
    "keysCorrect": 185,
    "keyAccuracy": 0.9536,
    "keyCi": [
     0.9142,
     0.9754
    ],
    "unsafeAllCases": 4
   }
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "jev",
   "routerLabel": "Jev 1.13 (TypeSafe)",
   "purpose": "control.failure_class",
   "purposeLabel": "Failure class",
   "arm": "jev",
   "calls": 18,
   "errors": 0,
   "cases": 18,
   "exact": 18,
   "exactRate": 1,
   "exactCi": [
    0.8241,
    1
   ],
   "keys": 18,
   "keysCorrect": 18,
   "keyAccuracy": 1,
   "keyCi": [
    0.8241,
    1
   ],
   "unsafeAllCases": 0,
   "questions": [
    {
     "key": "failure",
     "scored": 18,
     "correct": 18,
     "accuracy": 1,
     "abstention": 0.0556
    }
   ],
   "confusion": {
    "failure": {
     "impossible-here": {
      "impossible-here": 4
     },
     "flaky": {
      "flaky": 1
     },
     "real": {
      "real": 12,
      "none-of-these": 1
     }
    }
   },
   "costMicroUsd": 501
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "jev",
   "routerLabel": "Jev 1.13 (TypeSafe)",
   "purpose": "frontdoor.intent",
   "purposeLabel": "Message intent",
   "arm": "jev",
   "calls": 20,
   "errors": 0,
   "cases": 20,
   "exact": 20,
   "exactRate": 1,
   "exactCi": [
    0.8389,
    1
   ],
   "keys": 20,
   "keysCorrect": 20,
   "keyAccuracy": 1,
   "keyCi": [
    0.8389,
    1
   ],
   "unsafeAllCases": 0,
   "questions": [
    {
     "key": "intent",
     "scored": 20,
     "correct": 20,
     "accuracy": 1,
     "abstention": 0.1
    }
   ],
   "confusion": {
    "intent": {
     "request": {
      "request": 6
     },
     "status": {
      "status": 4
     },
     "followup": {
      "followup": 2
     },
     "decision": {
      "decision": 2
     },
     "control": {
      "control": 4
     },
     "none-of-these": {
      "none-of-these": 2
     }
    }
   },
   "costMicroUsd": 440
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "jev",
   "routerLabel": "Jev 1.13 (TypeSafe)",
   "purpose": "memory.is_rule",
   "purposeLabel": "Is it a rule?",
   "arm": "jev",
   "calls": 12,
   "errors": 0,
   "cases": 12,
   "exact": 12,
   "exactRate": 1,
   "exactCi": [
    0.7575,
    1
   ],
   "keys": 12,
   "keysCorrect": 12,
   "keyAccuracy": 1,
   "keyCi": [
    0.7575,
    1
   ],
   "unsafeAllCases": 0,
   "questions": [
    {
     "key": "kind",
     "scored": 12,
     "correct": 12,
     "accuracy": 1,
     "abstention": 0
    }
   ],
   "confusion": {
    "kind": {
     "not-a-rule": {
      "not-a-rule": 7,
      "obligation": 1
     },
     "obligation": {
      "obligation": 3
     },
     "prohibition": {
      "prohibition": 1
     }
    }
   },
   "costMicroUsd": 240
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "jev",
   "routerLabel": "Jev 1.13 (TypeSafe)",
   "purpose": "context.shape",
   "purposeLabel": "Context shape",
   "arm": "jev",
   "calls": 32,
   "errors": 0,
   "cases": 32,
   "exact": 24,
   "exactRate": 0.75,
   "exactCi": [
    0.5789,
    0.8675
   ],
   "keys": 144,
   "keysCorrect": 135,
   "keyAccuracy": 0.9375,
   "keyCi": [
    0.8855,
    0.9668
   ],
   "unsafeAllCases": 4,
   "questions": [
    {
     "key": "turn",
     "scored": 6,
     "correct": 5,
     "accuracy": 0.8333,
     "abstention": 0
    },
    {
     "key": "transcript",
     "scored": 16,
     "correct": 12,
     "accuracy": 0.75,
     "abstention": 0
    },
    {
     "key": "artifacts",
     "scored": 4,
     "correct": 4,
     "accuracy": 1,
     "abstention": 0
    },
    {
     "key": "knowledge",
     "scored": 24,
     "correct": 22,
     "accuracy": 0.9167,
     "abstention": 0
    },
    {
     "key": "memories",
     "scored": 21,
     "correct": 21,
     "accuracy": 1,
     "abstention": 0
    },
    {
     "key": "examples",
     "scored": 10,
     "correct": 10,
     "accuracy": 1,
     "abstention": 0
    },
    {
     "key": "scope",
     "scored": 31,
     "correct": 30,
     "accuracy": 0.9677,
     "abstention": 0
    },
    {
     "key": "complexity",
     "scored": 32,
     "correct": 31,
     "accuracy": 0.9688,
     "abstention": 0
    }
   ],
   "confusion": {
    "turn": {
     "follow-up": {
      "follow-up": 3
     },
     "clarification": {
      "new-task": 1,
      "clarification": 2
     }
    },
    "artifacts": {
     "no": {
      "no": 4
     }
    },
    "knowledge": {
     "search": {
      "search": 10
     },
     "none": {
      "facts": 1,
      "search": 5,
      "none": 1
     },
     "facts": {
      "search": 6,
      "facts": 1
     }
    },
    "memories": {
     "yes": {
      "yes": 11
     },
     "no": {
      "yes": 9,
      "no": 1
     }
    },
    "examples": {
     "yes": {
      "yes": 5
     },
     "no": {
      "no": 4,
      "yes": 1
     }
    },
    "scope": {
     "medium": {
      "medium": 11
     },
     "large": {
      "large": 5
     },
     "small": {
      "medium": 10,
      "small": 5
     }
    },
    "complexity": {
     "moderate": {
      "moderate": 20,
      "complex": 3
     },
     "complex": {
      "complex": 2,
      "moderate": 1
     },
     "simple": {
      "moderate": 4,
      "simple": 2
     }
    }
   },
   "costMicroUsd": 1584
  },
  {
   "batch": "batch-1",
   "type": "router",
   "id": "claude-haiku",
   "label": "Claude Haiku 4.5",
   "kind": "LLM via Claude Code CLI",
   "measured": "this benchmark, Claude Code CLI on a subscription account, one call per decision",
   "calls": 82,
   "okCalls": 82,
   "errorCalls": 0,
   "models": {
    "claude-haiku-4-5": 82
   },
   "latencyMs": {
    "apiP50": 10734,
    "apiP95": 32072,
    "wallP50": 12674,
    "wallP95": 34413,
    "cliOverheadP50": 1701
   },
   "tokensPerDecision": {
    "input": 1831.4,
    "cacheRead": 0,
    "cacheWrite": 0,
    "output": 1418.6,
    "thinking": 1100.8
   },
   "listCostPer1000Usd": 8.924,
   "cliReportedCostPer1000Usd": 8.924,
   "stopped": null,
   "costSource": "list price x CLI-reported tokens",
   "pooled": {
    "cases": 82,
    "exact": 73,
    "exactRate": 0.8902,
    "exactCi": [
     0.8044,
     0.9412
    ],
    "keys": 194,
    "keysCorrect": 183,
    "keyAccuracy": 0.9433,
    "keyCi": [
     0.9013,
     0.968
    ],
    "unsafeAllCases": 6
   }
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "claude-haiku",
   "routerLabel": "Claude Haiku 4.5",
   "purpose": "control.failure_class",
   "purposeLabel": "Failure class",
   "arm": "claude-haiku",
   "calls": 18,
   "errors": 0,
   "cases": 18,
   "exact": 17,
   "exactRate": 0.9444,
   "exactCi": [
    0.7424,
    0.9901
   ],
   "keys": 18,
   "keysCorrect": 17,
   "keyAccuracy": 0.9444,
   "keyCi": [
    0.7424,
    0.9901
   ],
   "unsafeAllCases": 1,
   "questions": [
    {
     "key": "failure",
     "scored": 18,
     "correct": 17,
     "accuracy": 0.9444,
     "abstention": 0.0556
    }
   ],
   "confusion": {
    "failure": {
     "impossible-here": {
      "impossible-here": 4
     },
     "flaky": {
      "flaky": 1
     },
     "real": {
      "real": 11,
      "flaky": 1,
      "none-of-these": 1
     }
    }
   },
   "costMicroUsd": 117165
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "claude-haiku",
   "routerLabel": "Claude Haiku 4.5",
   "purpose": "frontdoor.intent",
   "purposeLabel": "Message intent",
   "arm": "claude-haiku",
   "calls": 20,
   "errors": 0,
   "cases": 20,
   "exact": 20,
   "exactRate": 1,
   "exactCi": [
    0.8389,
    1
   ],
   "keys": 20,
   "keysCorrect": 20,
   "keyAccuracy": 1,
   "keyCi": [
    0.8389,
    1
   ],
   "unsafeAllCases": 0,
   "questions": [
    {
     "key": "intent",
     "scored": 20,
     "correct": 20,
     "accuracy": 1,
     "abstention": 0.1
    }
   ],
   "confusion": {
    "intent": {
     "request": {
      "request": 6
     },
     "status": {
      "status": 4
     },
     "followup": {
      "followup": 2
     },
     "decision": {
      "decision": 2
     },
     "control": {
      "control": 4
     },
     "none-of-these": {
      "none-of-these": 2
     }
    }
   },
   "costMicroUsd": 94066
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "claude-haiku",
   "routerLabel": "Claude Haiku 4.5",
   "purpose": "memory.is_rule",
   "purposeLabel": "Is it a rule?",
   "arm": "claude-haiku",
   "calls": 12,
   "errors": 0,
   "cases": 12,
   "exact": 12,
   "exactRate": 1,
   "exactCi": [
    0.7575,
    1
   ],
   "keys": 12,
   "keysCorrect": 12,
   "keyAccuracy": 1,
   "keyCi": [
    0.7575,
    1
   ],
   "unsafeAllCases": 0,
   "questions": [
    {
     "key": "kind",
     "scored": 12,
     "correct": 12,
     "accuracy": 1,
     "abstention": 0
    }
   ],
   "confusion": {
    "kind": {
     "not-a-rule": {
      "not-a-rule": 7,
      "obligation": 1
     },
     "obligation": {
      "not-a-rule": 1,
      "prohibition": 2
     },
     "prohibition": {
      "prohibition": 1
     }
    }
   },
   "costMicroUsd": 94777
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "claude-haiku",
   "routerLabel": "Claude Haiku 4.5",
   "purpose": "context.shape",
   "purposeLabel": "Context shape",
   "arm": "claude-haiku",
   "calls": 32,
   "errors": 0,
   "cases": 32,
   "exact": 24,
   "exactRate": 0.75,
   "exactCi": [
    0.5789,
    0.8675
   ],
   "keys": 144,
   "keysCorrect": 134,
   "keyAccuracy": 0.9306,
   "keyCi": [
    0.8769,
    0.9618
   ],
   "unsafeAllCases": 5,
   "questions": [
    {
     "key": "turn",
     "scored": 6,
     "correct": 5,
     "accuracy": 0.8333,
     "abstention": 0
    },
    {
     "key": "transcript",
     "scored": 16,
     "correct": 13,
     "accuracy": 0.8125,
     "abstention": 0
    },
    {
     "key": "artifacts",
     "scored": 4,
     "correct": 4,
     "accuracy": 1,
     "abstention": 0
    },
    {
     "key": "knowledge",
     "scored": 24,
     "correct": 22,
     "accuracy": 0.9167,
     "abstention": 0
    },
    {
     "key": "memories",
     "scored": 21,
     "correct": 21,
     "accuracy": 1,
     "abstention": 0
    },
    {
     "key": "examples",
     "scored": 10,
     "correct": 10,
     "accuracy": 1,
     "abstention": 0
    },
    {
     "key": "scope",
     "scored": 31,
     "correct": 29,
     "accuracy": 0.9355,
     "abstention": 0
    },
    {
     "key": "complexity",
     "scored": 32,
     "correct": 30,
     "accuracy": 0.9375,
     "abstention": 0
    }
   ],
   "confusion": {
    "turn": {
     "follow-up": {
      "follow-up": 3
     },
     "clarification": {
      "follow-up": 1,
      "clarification": 2
     }
    },
    "artifacts": {
     "no": {
      "yes": 3,
      "no": 1
     }
    },
    "knowledge": {
     "search": {
      "search": 10
     },
     "none": {
      "search": 6,
      "facts": 1
     },
     "facts": {
      "search": 7
     }
    },
    "memories": {
     "yes": {
      "yes": 11
     },
     "no": {
      "yes": 8,
      "no": 2
     }
    },
    "examples": {
     "yes": {
      "yes": 5
     },
     "no": {
      "yes": 4,
      "no": 1
     }
    },
    "scope": {
     "medium": {
      "medium": 10,
      "small": 1
     },
     "large": {
      "large": 5
     },
     "small": {
      "medium": 9,
      "small": 6
     }
    },
    "complexity": {
     "moderate": {
      "moderate": 18,
      "complex": 3,
      "simple": 2
     },
     "complex": {
      "complex": 3
     },
     "simple": {
      "simple": 3,
      "moderate": 3
     }
    }
   },
   "costMicroUsd": 425793
  },
  {
   "batch": "batch-1",
   "type": "router",
   "id": "claude-sonnet",
   "label": "Claude Sonnet 5.5",
   "kind": "LLM via Claude Code CLI",
   "measured": "this benchmark, Claude Code CLI on a subscription account, one call per decision",
   "calls": 82,
   "okCalls": 82,
   "errorCalls": 0,
   "models": {
    "claude-sonnet-5-5": 82
   },
   "latencyMs": {
    "apiP50": 1599,
    "apiP95": 2574,
    "wallP50": 2598,
    "wallP95": 4298,
    "cliOverheadP50": 975
   },
   "tokensPerDecision": {
    "input": 2,
    "cacheRead": 230.8,
    "cacheWrite": 1551.9,
    "output": 106.6,
    "thinking": 1.5
   },
   "listCostPer1000Usd": 4.996,
   "cliReportedCostPer1000Usd": 7.324,
   "stopped": null,
   "costSource": "list price x CLI-reported tokens",
   "pooled": {
    "cases": 82,
    "exact": 77,
    "exactRate": 0.939,
    "exactCi": [
     0.8651,
     0.9737
    ],
    "keys": 194,
    "keysCorrect": 189,
    "keyAccuracy": 0.9742,
    "keyCi": [
     0.9411,
     0.9889
    ],
    "unsafeAllCases": 2
   }
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "claude-sonnet",
   "routerLabel": "Claude Sonnet 5.5",
   "purpose": "control.failure_class",
   "purposeLabel": "Failure class",
   "arm": "claude-sonnet",
   "calls": 18,
   "errors": 0,
   "cases": 18,
   "exact": 18,
   "exactRate": 1,
   "exactCi": [
    0.8241,
    1
   ],
   "keys": 18,
   "keysCorrect": 18,
   "keyAccuracy": 1,
   "keyCi": [
    0.8241,
    1
   ],
   "unsafeAllCases": 0,
   "questions": [
    {
     "key": "failure",
     "scored": 18,
     "correct": 18,
     "accuracy": 1,
     "abstention": 0.0556
    }
   ],
   "confusion": {
    "failure": {
     "impossible-here": {
      "impossible-here": 4
     },
     "flaky": {
      "flaky": 1
     },
     "real": {
      "real": 12,
      "none-of-these": 1
     }
    }
   },
   "costMicroUsd": 125938
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "claude-sonnet",
   "routerLabel": "Claude Sonnet 5.5",
   "purpose": "frontdoor.intent",
   "purposeLabel": "Message intent",
   "arm": "claude-sonnet",
   "calls": 20,
   "errors": 0,
   "cases": 20,
   "exact": 20,
   "exactRate": 1,
   "exactCi": [
    0.8389,
    1
   ],
   "keys": 20,
   "keysCorrect": 20,
   "keyAccuracy": 1,
   "keyCi": [
    0.8389,
    1
   ],
   "unsafeAllCases": 0,
   "questions": [
    {
     "key": "intent",
     "scored": 20,
     "correct": 20,
     "accuracy": 1,
     "abstention": 0.1
    }
   ],
   "confusion": {
    "intent": {
     "request": {
      "request": 6
     },
     "status": {
      "status": 4
     },
     "followup": {
      "followup": 2
     },
     "decision": {
      "decision": 2
     },
     "control": {
      "control": 4
     },
     "none-of-these": {
      "none-of-these": 2
     }
    }
   },
   "costMicroUsd": 122632
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "claude-sonnet",
   "routerLabel": "Claude Sonnet 5.5",
   "purpose": "memory.is_rule",
   "purposeLabel": "Is it a rule?",
   "arm": "claude-sonnet",
   "calls": 12,
   "errors": 0,
   "cases": 12,
   "exact": 12,
   "exactRate": 1,
   "exactCi": [
    0.7575,
    1
   ],
   "keys": 12,
   "keysCorrect": 12,
   "keyAccuracy": 1,
   "keyCi": [
    0.7575,
    1
   ],
   "unsafeAllCases": 0,
   "questions": [
    {
     "key": "kind",
     "scored": 12,
     "correct": 12,
     "accuracy": 1,
     "abstention": 0
    }
   ],
   "confusion": {
    "kind": {
     "not-a-rule": {
      "not-a-rule": 7,
      "obligation": 1
     },
     "obligation": {
      "obligation": 2,
      "prohibition": 1
     },
     "prohibition": {
      "prohibition": 1
     }
    }
   },
   "costMicroUsd": 74846
  },
  {
   "batch": "batch-1",
   "type": "router-purpose",
   "router": "claude-sonnet",
   "routerLabel": "Claude Sonnet 5.5",
   "purpose": "context.shape",
   "purposeLabel": "Context shape",
   "arm": "claude-sonnet",
   "calls": 32,
   "errors": 0,
   "cases": 32,
   "exact": 27,
   "exactRate": 0.8438,
   "exactCi": [
    0.6825,
    0.9314
   ],
   "keys": 144,
   "keysCorrect": 139,
   "keyAccuracy": 0.9653,
   "keyCi": [
    0.9213,
    0.9851
   ],
   "unsafeAllCases": 2,
   "questions": [
    {
     "key": "turn",
     "scored": 6,
     "correct": 5,
     "accuracy": 0.8333,
     "abstention": 0
    },
    {
     "key": "transcript",
     "scored": 16,
     "correct": 15,
     "accuracy": 0.9375,
     "abstention": 0
    },
    {
     "key": "artifacts",
     "scored": 4,
     "correct": 4,
     "accuracy": 1,
     "abstention": 0
    },
    {
     "key": "knowledge",
     "scored": 24,
     "correct": 23,
     "accuracy": 0.9583,
     "abstention": 0
    },
    {
     "key": "memories",
     "scored": 21,
     "correct": 20,
     "accuracy": 0.9524,
     "abstention": 0
    },
    {
     "key": "examples",
     "scored": 10,
     "correct": 10,
     "accuracy": 1,
     "abstention": 0
    },
    {
     "key": "scope",
     "scored": 31,
     "correct": 31,
     "accuracy": 1,
     "abstention": 0
    },
    {
     "key": "complexity",
     "scored": 32,
     "correct": 31,
     "accuracy": 0.9688,
     "abstention": 0
    }
   ],
   "confusion": {
    "turn": {
     "follow-up": {
      "follow-up": 3
     },
     "clarification": {
      "follow-up": 1,
      "clarification": 2
     }
    },
    "artifacts": {
     "no": {
      "no": 3,
      "yes": 1
     }
    },
    "knowledge": {
     "search": {
      "search": 10
     },
     "none": {
      "search": 5,
      "facts": 2
     },
     "facts": {
      "search": 5,
      "facts": 2
     }
    },
    "memories": {
     "yes": {
      "yes": 11
     },
     "no": {
      "yes": 10
     }
    },
    "examples": {
     "yes": {
      "yes": 5
     },
     "no": {
      "yes": 3,
      "no": 2
     }
    },
    "scope": {
     "medium": {
      "medium": 11
     },
     "large": {
      "large": 5
     },
     "small": {
      "medium": 8,
      "small": 7
     }
    },
    "complexity": {
     "moderate": {
      "moderate": 20,
      "complex": 3
     },
     "complex": {
      "complex": 2,
      "moderate": 1
     },
     "simple": {
      "moderate": 2,
      "simple": 4
     }
    }
   },
   "costMicroUsd": 277131
  },
  {
   "batch": "batch-1",
   "type": "router",
   "id": "clef",
   "label": "Clef / Clef-Flash (local)",
   "kind": "decision model",
   "measured": null,
   "notMeasuredReason": "No local Clef server was running and installing a 6-20 GB model was out of scope for this run.",
   "pooled": null
  },
  {
   "batch": "batch-1",
   "type": "head-to-head",
   "router": "claude-haiku",
   "versus": "jev",
   "bothRight": 70,
   "onlyJev": 4,
   "onlyRouter": 3,
   "bothWrong": 5,
   "cases": [
    {
     "purpose": "control.failure_class",
     "case": "own-server-refused",
     "jev": true,
     "router": false
    },
    {
     "purpose": "context.shape",
     "case": "new-design-decision",
     "jev": false,
     "router": true
    },
    {
     "purpose": "context.shape",
     "case": "ml-replan-oversized",
     "jev": true,
     "router": false
    },
    {
     "purpose": "context.shape",
     "case": "ml-cont-failing-suite",
     "jev": false,
     "router": true
    },
    {
     "purpose": "context.shape",
     "case": "ml-new-changelog-list",
     "jev": true,
     "router": false
    },
    {
     "purpose": "context.shape",
     "case": "asked-followup-csv",
     "jev": true,
     "router": false
    },
    {
     "purpose": "context.shape",
     "case": "asked-clarify-limit",
     "jev": false,
     "router": true
    }
   ],
   "mcnemarExactP": 1
  },
  {
   "batch": "batch-1",
   "type": "head-to-head",
   "router": "claude-sonnet",
   "versus": "jev",
   "bothRight": 73,
   "onlyJev": 1,
   "onlyRouter": 4,
   "bothWrong": 4,
   "cases": [
    {
     "purpose": "context.shape",
     "case": "ml-cont-failing-suite",
     "jev": false,
     "router": true
    },
    {
     "purpose": "context.shape",
     "case": "ml-new-changelog-list",
     "jev": true,
     "router": false
    },
    {
     "purpose": "context.shape",
     "case": "asked-clarify-limit",
     "jev": false,
     "router": true
    },
    {
     "purpose": "context.shape",
     "case": "asked-followup-notes",
     "jev": false,
     "router": true
    },
    {
     "purpose": "context.shape",
     "case": "asked-clarify-branch",
     "jev": false,
     "router": true
    }
   ],
   "mcnemarExactP": 0.375
  },
  {
   "batch": "batch-1",
   "type": "economics-scenario",
   "scenario": "all Haiku 4.5",
   "usd": 54.2715,
   "vsAllSonnet": 0.5,
   "runs": 50,
   "calls": 2362
  },
  {
   "batch": "batch-1",
   "type": "economics-scenario",
   "scenario": "all Sonnet 5.5",
   "usd": 108.543,
   "vsAllSonnet": 1,
   "runs": 50,
   "calls": 2362
  },
  {
   "batch": "batch-1",
   "type": "economics-scenario",
   "scenario": "all Opus 5.5",
   "usd": 170.916,
   "vsAllSonnet": 1.5746,
   "runs": 50,
   "calls": 2362
  },
  {
   "batch": "batch-1",
   "type": "economics-scenario",
   "scenario": "all Fable 5.1",
   "usd": 369.5774,
   "vsAllSonnet": 3.4049,
   "runs": 50,
   "calls": 2362
  },
  {
   "batch": "batch-1",
   "type": "economics-scenario",
   "scenario": "policy (Opus strong, Haiku ancillary)",
   "usd": 161.6161,
   "vsAllSonnet": 1.489,
   "runs": 50,
   "calls": 2362
  },
  {
   "batch": "batch-1",
   "type": "economics-scenario",
   "scenario": "split (Sonnet main line, Haiku ancillary)",
   "usd": 105.5331,
   "vsAllSonnet": 0.9723,
   "runs": 50,
   "calls": 2362
  },
  {
   "batch": "batch-1",
   "type": "economics-stage",
   "stage": "act",
   "calls": 1320,
   "inputTokens": 185219212,
   "cachedTokens": 178822501,
   "outputTokens": 875211,
   "callsOver200k": 1,
   "tier": "strong",
   "cost": {
    "Haiku 4.5": 28.6550161,
    "Sonnet 5.5": 57.3100322,
    "Opus 5.5": 78.8555642,
    "Fable 5.1": 152.43328525
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-stage",
   "stage": "research",
   "calls": 445,
   "inputTokens": 55440296,
   "cachedTokens": 50049026,
   "outputTokens": 783098,
   "callsOver200k": 1,
   "tier": "strong",
   "cost": {
    "Haiku 4.5": 14.311662600000002,
    "Sonnet 5.5": 28.623325200000004,
    "Opus 5.5": 47.236845200000005,
    "Fable 5.1": 105.5798565
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-stage",
   "stage": "verify",
   "calls": 105,
   "inputTokens": 2984655,
   "cachedTokens": 303588,
   "outputTokens": 405850,
   "callsOver200k": 0,
   "tier": "strong",
   "cost": {
    "Haiku 4.5": 4.7406758,
    "Sonnet 5.5": 9.4813516,
    "Opus 5.5": 18.9019856,
    "Fable 5.1": 47.179067
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-stage",
   "stage": "review",
   "calls": 140,
   "inputTokens": 2269371,
   "cachedTokens": 1034443,
   "outputTokens": 403349,
   "callsOver200k": 0,
   "tier": "strong",
   "cost": {
    "Haiku 4.5": 3.3551173,
    "Sonnet 5.5": 6.7102346,
    "Opus 5.5": 13.2135806,
    "Fable 5.1": 32.77534075
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-stage",
   "stage": "memory and onboarding",
   "calls": 337,
   "inputTokens": 2134225,
   "cachedTokens": 629530,
   "outputTokens": 308897,
   "callsOver200k": 0,
   "tier": "economy",
   "cost": {
    "Haiku 4.5": 3.112133,
    "Sonnet 5.5": 6.224266,
    "Opus 5.5": 12.322626,
    "Fable 5.1": 30.6491825
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-stage",
   "stage": "other",
   "calls": 15,
   "inputTokens": 80089,
   "cachedTokens": 10923,
   "outputTokens": 5326,
   "callsOver200k": 0,
   "tier": "standard",
   "cost": {
    "Haiku 4.5": 0.0968883,
    "Sonnet 5.5": 0.1937766,
    "Opus 5.5": 0.38536859999999995,
    "Fable 5.1": 0.96069075
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "astropy_astropy_13579_1",
   "calls": 62,
   "inputTokens": 6954584,
   "cachedTokens": 6536881,
   "outputTokens": 78229,
   "recordedCostUsd": 3.760223,
   "cost": {
    "Haiku 4.5": 1.4625361000000001,
    "Sonnet 5.5": 2.9250722000000002,
    "Opus 5.5": 4.5427682,
    "Fable 5.1": 9.72270025,
    "policy (Opus strong, Haiku ancillary)": 4.324064399999999,
    "split (Sonnet main line, Haiku ancillary)": 2.8509584
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "astropy_astropy_14096_1",
   "calls": 62,
   "inputTokens": 6580882,
   "cachedTokens": 6252311,
   "outputTokens": 60668,
   "recordedCostUsd": 3.1711750000000003,
   "cost": {
    "Haiku 4.5": 1.2571421,
    "Sonnet 5.5": 2.5142842,
    "Opus 5.5": 3.7781062000000003,
    "Fable 5.1": 7.88218775,
    "policy (Opus strong, Haiku ancillary)": 3.5515897,
    "split (Sonnet main line, Haiku ancillary)": 2.4377137
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "astropy_astropy_7336_1",
   "calls": 47,
   "inputTokens": 4971964,
   "cachedTokens": 4803398,
   "outputTokens": 30879,
   "recordedCostUsd": 1.9435399999999998,
   "cost": {
    "Haiku 4.5": 0.8033008,
    "Sonnet 5.5": 1.6066016,
    "Opus 5.5": 2.2525236,
    "Fable 5.1": 4.4304595,
    "policy (Opus strong, Haiku ancillary)": 2.0814224,
    "split (Sonnet main line, Haiku ancillary)": 1.5485584000000001
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_10554_1",
   "calls": 33,
   "inputTokens": 3216448,
   "cachedTokens": 3084636,
   "outputTokens": 28058,
   "recordedCostUsd": 1.424618,
   "cost": {
    "Haiku 4.5": 0.5805656000000001,
    "Sonnet 5.5": 1.1611312000000003,
    "Opus 5.5": 1.7053352000000002,
    "Fable 5.1": 3.492179,
    "policy (Opus strong, Haiku ancillary)": 1.5314045,
    "split (Sonnet main line, Haiku ancillary)": 1.1050945
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_11333_1",
   "calls": 45,
   "inputTokens": 4637892,
   "cachedTokens": 4396072,
   "outputTokens": 55592,
   "recordedCostUsd": 2.4022309999999996,
   "cost": {
    "Haiku 4.5": 0.9593872,
    "Sonnet 5.5": 1.9187744,
    "Opus 5.5": 2.9583344,
    "Fable 5.1": 6.296818,
    "policy (Opus strong, Haiku ancillary)": 2.8096658999999997,
    "split (Sonnet main line, Haiku ancillary)": 1.8683359
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_11885_1",
   "calls": 41,
   "inputTokens": 4300260,
   "cachedTokens": 3928137,
   "outputTokens": 59863,
   "recordedCostUsd": 2.8725799999999997,
   "cost": {
    "Haiku 4.5": 1.0642517,
    "Sonnet 5.5": 2.1285034,
    "Opus 5.5": 3.4713794,
    "Fable 5.1": 7.69641425,
    "policy (Opus strong, Haiku ancillary)": 3.2753702999999996,
    "split (Sonnet main line, Haiku ancillary)": 2.0624023
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_11885_2",
   "calls": 36,
   "inputTokens": 3925238,
   "cachedTokens": 3713701,
   "outputTokens": 47485,
   "recordedCostUsd": 2.06359,
   "cost": {
    "Haiku 4.5": 0.8203321000000001,
    "Sonnet 5.5": 1.6406642000000002,
    "Opus 5.5": 2.5385882000000004,
    "Fable 5.1": 5.41804525,
    "policy (Opus strong, Haiku ancillary)": 2.3668353,
    "split (Sonnet main line, Haiku ancillary)": 1.5824673
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_12193_1",
   "calls": 42,
   "inputTokens": 4076373,
   "cachedTokens": 3862815,
   "outputTokens": 36215,
   "recordedCostUsd": 1.988769,
   "cost": {
    "Haiku 4.5": 0.7809145,
    "Sonnet 5.5": 1.561829,
    "Opus 5.5": 2.351095,
    "Fable 5.1": 4.91203375,
    "policy (Opus strong, Haiku ancillary)": 2.2042956000000005,
    "split (Sonnet main line, Haiku ancillary)": 1.5119496000000001
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_12741_1",
   "calls": 52,
   "inputTokens": 5557408,
   "cachedTokens": 5273734,
   "outputTokens": 62870,
   "recordedCostUsd": 2.817931,
   "cost": {
    "Haiku 4.5": 1.1253974,
    "Sonnet 5.5": 2.2507948,
    "Opus 5.5": 3.4468427999999998,
    "Fable 5.1": 7.2986735,
    "policy (Opus strong, Haiku ancillary)": 3.2959294,
    "split (Sonnet main line, Haiku ancillary)": 2.1995514000000003
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_13158_1",
   "calls": 42,
   "inputTokens": 3915952,
   "cachedTokens": 3693270,
   "outputTokens": 43980,
   "recordedCostUsd": 2.0690069999999996,
   "cost": {
    "Haiku 4.5": 0.811909,
    "Sonnet 5.5": 1.623818,
    "Opus 5.5": 2.508982,
    "Fable 5.1": 5.3491375,
    "policy (Opus strong, Haiku ancillary)": 2.3233459,
    "split (Sonnet main line, Haiku ancillary)": 1.5741359
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_13925_1",
   "calls": 42,
   "inputTokens": 3974666,
   "cachedTokens": 3713559,
   "outputTokens": 42385,
   "recordedCostUsd": 2.2108149999999998,
   "cost": {
    "Haiku 4.5": 0.8443879,
    "Sonnet 5.5": 1.6887758,
    "Opus 5.5": 2.6348398,
    "Fable 5.1": 5.65870975,
    "policy (Opus strong, Haiku ancillary)": 2.4560575,
    "split (Sonnet main line, Haiku ancillary)": 1.6281094999999999
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_14034_1",
   "calls": 52,
   "inputTokens": 5107100,
   "cachedTokens": 4808603,
   "outputTokens": 65148,
   "recordedCostUsd": 2.8069789999999997,
   "cost": {
    "Haiku 4.5": 1.1050973,
    "Sonnet 5.5": 2.2101946,
    "Opus 5.5": 3.4586686,
    "Fable 5.1": 7.44452075,
    "policy (Opus strong, Haiku ancillary)": 3.2861172999999995,
    "split (Sonnet main line, Haiku ancillary)": 2.1522073
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_14373_1",
   "calls": 37,
   "inputTokens": 3155048,
   "cachedTokens": 2978619,
   "outputTokens": 35113,
   "recordedCostUsd": 1.652423,
   "cost": {
    "Haiku 4.5": 0.6498559,
    "Sonnet 5.5": 1.2997118,
    "Opus 5.5": 2.0036998,
    "Fable 5.1": 4.26459475,
    "policy (Opus strong, Haiku ancillary)": 1.8249254999999998,
    "split (Sonnet main line, Haiku ancillary)": 1.2424834999999999
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_14559_1",
   "calls": 46,
   "inputTokens": 4478121,
   "cachedTokens": 4137430,
   "outputTokens": 42645,
   "recordedCostUsd": 2.6165130000000003,
   "cost": {
    "Haiku 4.5": 0.967659,
    "Sonnet 5.5": 1.935318,
    "Opus 5.5": 3.04315,
    "Fable 5.1": 6.5735175,
    "policy (Opus strong, Haiku ancillary)": 2.8569070000000005,
    "split (Sonnet main line, Haiku ancillary)": 1.8722310000000002
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_14631_1",
   "calls": 45,
   "inputTokens": 4895047,
   "cachedTokens": 4620790,
   "outputTokens": 60006,
   "recordedCostUsd": 2.621057,
   "cost": {
    "Haiku 4.5": 1.036366,
    "Sonnet 5.5": 2.072732,
    "Opus 5.5": 3.221306,
    "Fable 5.1": 6.8980675,
    "policy (Opus strong, Haiku ancillary)": 3.04635,
    "split (Sonnet main line, Haiku ancillary)": 2.01341
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_14672_1",
   "calls": 56,
   "inputTokens": 5797649,
   "cachedTokens": 5502159,
   "outputTokens": 55768,
   "recordedCostUsd": 2.839842,
   "cost": {
    "Haiku 4.5": 1.1245459,
    "Sonnet 5.5": 2.2490918,
    "Opus 5.5": 3.3977518,
    "Fable 5.1": 7.11883975,
    "policy (Opus strong, Haiku ancillary)": 3.1903088,
    "split (Sonnet main line, Haiku ancillary)": 2.1879848
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_15022_1",
   "calls": 59,
   "inputTokens": 6399305,
   "cachedTokens": 5997010,
   "outputTokens": 94182,
   "recordedCostUsd": 3.7501569999999997,
   "cost": {
    "Haiku 4.5": 1.472906,
    "Sonnet 5.5": 2.945812,
    "Opus 5.5": 4.692222,
    "Fable 5.1": 10.2313025,
    "policy (Opus strong, Haiku ancillary)": 4.4949494,
    "split (Sonnet main line, Haiku ancillary)": 2.8790514
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_15280_1",
   "calls": 56,
   "inputTokens": 6152134,
   "cachedTokens": 5768910,
   "outputTokens": 68186,
   "recordedCostUsd": 3.3683050000000003,
   "cost": {
    "Haiku 4.5": 1.301045,
    "Sonnet 5.5": 2.60209,
    "Opus 5.5": 4.050398,
    "Fable 5.1": 8.6837675,
    "policy (Opus strong, Haiku ancillary)": 3.8714290000000005,
    "split (Sonnet main line, Haiku ancillary)": 2.541429
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_15572_1",
   "calls": 41,
   "inputTokens": 3701457,
   "cachedTokens": 3529153,
   "outputTokens": 33978,
   "recordedCostUsd": 1.7346599999999999,
   "cost": {
    "Haiku 4.5": 0.6951093,
    "Sonnet 5.5": 1.3902186,
    "Opus 5.5": 2.0746066,
    "Fable 5.1": 4.30422825,
    "policy (Opus strong, Haiku ancillary)": 1.8637894,
    "split (Sonnet main line, Haiku ancillary)": 1.3222254
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_15732_1",
   "calls": 57,
   "inputTokens": 5968541,
   "cachedTokens": 5688439,
   "outputTokens": 48017,
   "recordedCostUsd": 2.7380329999999997,
   "cost": {
    "Haiku 4.5": 1.0890308999999998,
    "Sonnet 5.5": 2.1780617999999996,
    "Opus 5.5": 3.2184358,
    "Fable 5.1": 6.62397975,
    "policy (Opus strong, Haiku ancillary)": 3.0310521,
    "split (Sonnet main line, Haiku ancillary)": 2.1146061
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_16255_1",
   "calls": 45,
   "inputTokens": 4410098,
   "cachedTokens": 4132843,
   "outputTokens": 44373,
   "recordedCostUsd": 2.379139,
   "cost": {
    "Haiku 4.5": 0.9124043000000001,
    "Sonnet 5.5": 1.8248086000000001,
    "Opus 5.5": 2.8230486,
    "Fable 5.1": 6.02441075,
    "policy (Opus strong, Haiku ancillary)": 2.6545983,
    "split (Sonnet main line, Haiku ancillary)": 1.7681882999999998
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_16333_1",
   "calls": 36,
   "inputTokens": 3698413,
   "cachedTokens": 3557433,
   "outputTokens": 26113,
   "recordedCostUsd": 1.5363879999999999,
   "cost": {
    "Haiku 4.5": 0.6272883,
    "Sonnet 5.5": 1.2545766,
    "Opus 5.5": 1.7976666000000001,
    "Fable 5.1": 3.60480825,
    "policy (Opus strong, Haiku ancillary)": 1.6175646,
    "split (Sonnet main line, Haiku ancillary)": 1.1935386
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "django_django_17029_1",
   "calls": 41,
   "inputTokens": 3929377,
   "cachedTokens": 3680380,
   "outputTokens": 45033,
   "recordedCostUsd": 2.1822209999999997,
   "cost": {
    "Haiku 4.5": 0.8422,
    "Sonnet 5.5": 1.6844,
    "Opus 5.5": 2.632724,
    "Fable 5.1": 5.661715,
    "policy (Opus strong, Haiku ancillary)": 2.448743,
    "split (Sonnet main line, Haiku ancillary)": 1.6219989999999997
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "h3_pr1533_1",
   "calls": 44,
   "inputTokens": 4844162,
   "cachedTokens": 4451900,
   "outputTokens": 57377,
   "recordedCostUsd": 3.033016,
   "cost": {
    "Haiku 4.5": 1.124337,
    "Sonnet 5.5": 2.248674,
    "Opus 5.5": 3.606968,
    "Fable 5.1": 7.904445,
    "policy (Opus strong, Haiku ancillary)": 3.3707572999999997,
    "split (Sonnet main line, Haiku ancillary)": 2.1761872999999996
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "h3_pr1533_3",
   "calls": 50,
   "inputTokens": 5121396,
   "cachedTokens": 4599446,
   "outputTokens": 69182,
   "recordedCostUsd": 3.6993090000000004,
   "cost": {
    "Haiku 4.5": 1.3278046000000001,
    "Sonnet 5.5": 2.6556092000000002,
    "Opus 5.5": 4.3913292,
    "Fable 5.1": 9.8284615,
    "policy (Opus strong, Haiku ancillary)": 4.1345912,
    "split (Sonnet main line, Haiku ancillary)": 2.5761852
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "h3_pr1533_4",
   "calls": 45,
   "inputTokens": 4780204,
   "cachedTokens": 4249232,
   "outputTokens": 55844,
   "recordedCostUsd": 3.531988,
   "cost": {
    "Haiku 4.5": 1.2351151999999999,
    "Sonnet 5.5": 2.4702303999999997,
    "Opus 5.5": 4.0906144,
    "Fable 5.1": 9.164228,
    "policy (Opus strong, Haiku ancillary)": 3.8561607,
    "split (Sonnet main line, Haiku ancillary)": 2.3969627
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "h3_pr1533_5",
   "calls": 32,
   "inputTokens": 3043888,
   "cachedTokens": 2642586,
   "outputTokens": 49289,
   "recordedCostUsd": 2.62648,
   "cost": {
    "Haiku 4.5": 0.9120056000000001,
    "Sonnet 5.5": 1.8240112000000002,
    "Opus 5.5": 3.1195052000000003,
    "Fable 5.1": 7.1381165,
    "policy (Opus strong, Haiku ancillary)": 2.8125497,
    "split (Sonnet main line, Haiku ancillary)": 1.7411357
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "matplotlib_matplotlib_21568_1",
   "calls": 52,
   "inputTokens": 5528274,
   "cachedTokens": 5180677,
   "outputTokens": 52118,
   "recordedCostUsd": 2.947492,
   "cost": {
    "Haiku 4.5": 1.1262547,
    "Sonnet 5.5": 2.2525094,
    "Opus 5.5": 3.4688833999999997,
    "Fable 5.1": 7.37703925,
    "policy (Opus strong, Haiku ancillary)": 3.2809617000000006,
    "split (Sonnet main line, Haiku ancillary)": 2.1889217
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "matplotlib_matplotlib_24149_1",
   "calls": 62,
   "inputTokens": 6556359,
   "cachedTokens": 6179436,
   "outputTokens": 73977,
   "recordedCostUsd": 3.483097,
   "cost": {
    "Haiku 4.5": 1.3647516000000002,
    "Sonnet 5.5": 2.7295032000000004,
    "Opus 5.5": 4.2231192,
    "Fable 5.1": 9.012939,
    "policy (Opus strong, Haiku ancillary)": 4.0483622,
    "split (Sonnet main line, Haiku ancillary)": 2.6702941999999994
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "matplotlib_matplotlib_24970_1",
   "calls": 63,
   "inputTokens": 6942770,
   "cachedTokens": 6579879,
   "outputTokens": 86189,
   "recordedCostUsd": 3.629167,
   "cost": {
    "Haiku 4.5": 1.4518239,
    "Sonnet 5.5": 2.9036478,
    "Opus 5.5": 4.491319799999999,
    "Fable 5.1": 9.58332975,
    "policy (Opus strong, Haiku ancillary)": 4.314057,
    "split (Sonnet main line, Haiku ancillary)": 2.8435230000000002
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "pydata_xarray_6721_1",
   "calls": 54,
   "inputTokens": 6062262,
   "cachedTokens": 5556978,
   "outputTokens": 67600,
   "recordedCostUsd": 3.808312,
   "cost": {
    "Haiku 4.5": 1.3989818,
    "Sonnet 5.5": 2.7979636,
    "Opus 5.5": 4.4845315999999995,
    "Fable 5.1": 9.8220845,
    "policy (Opus strong, Haiku ancillary)": 4.3241053,
    "split (Sonnet main line, Haiku ancillary)": 2.7434593
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "scikit_learn_scikit_learn_14894_1",
   "calls": 48,
   "inputTokens": 5083893,
   "cachedTokens": 4683017,
   "outputTokens": 51659,
   "recordedCostUsd": 3.056502,
   "cost": {
    "Haiku 4.5": 1.1274727,
    "Sonnet 5.5": 2.2549454,
    "Opus 5.5": 3.5732874,
    "Fable 5.1": 7.76246425,
    "policy (Opus strong, Haiku ancillary)": 3.4429441,
    "split (Sonnet main line, Haiku ancillary)": 2.2166061
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "scikit_learn_scikit_learn_25973_1",
   "calls": 42,
   "inputTokens": 4364219,
   "cachedTokens": 4132922,
   "outputTokens": 46977,
   "recordedCostUsd": 2.2213700000000003,
   "cost": {
    "Haiku 4.5": 0.8794742,
    "Sonnet 5.5": 1.7589484,
    "Opus 5.5": 2.6913123999999997,
    "Fable 5.1": 5.6950505,
    "policy (Opus strong, Haiku ancillary)": 2.5423969000000004,
    "split (Sonnet main line, Haiku ancillary)": 1.7082989
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "session_pr348_1",
   "calls": 69,
   "inputTokens": 7572411,
   "cachedTokens": 6986001,
   "outputTokens": 71156,
   "recordedCostUsd": 4.454123,
   "cost": {
    "Haiku 4.5": 1.6407901,
    "Sonnet 5.5": 3.2815802,
    "Opus 5.5": 5.1659602,
    "Fable 5.1": 11.16840025,
    "policy (Opus strong, Haiku ancillary)": 4.984136100000001,
    "split (Sonnet main line, Haiku ancillary)": 3.2225801
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "session_pr348_2",
   "calls": 68,
   "inputTokens": 7280943,
   "cachedTokens": 6667133,
   "outputTokens": 84879,
   "recordedCostUsd": 4.637178999999999,
   "cost": {
    "Haiku 4.5": 1.7049183,
    "Sonnet 5.5": 3.4098366,
    "Opus 5.5": 5.486246599999999,
    "Fable 5.1": 12.04883325,
    "policy (Opus strong, Haiku ancillary)": 5.325332800000001,
    "split (Sonnet main line, Haiku ancillary)": 3.3582008000000005
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "session_pr348_3",
   "calls": 41,
   "inputTokens": 4183787,
   "cachedTokens": 3694031,
   "outputTokens": 54468,
   "recordedCostUsd": 3.242336,
   "cost": {
    "Haiku 4.5": 1.1314991,
    "Sonnet 5.5": 2.2629982,
    "Opus 5.5": 3.7871902000000004,
    "Fable 5.1": 8.54446775,
    "policy (Opus strong, Haiku ancillary)": 3.6149861000000003,
    "split (Sonnet main line, Haiku ancillary)": 2.2075081
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "session_pr348_4",
   "calls": 27,
   "inputTokens": 2547130,
   "cachedTokens": 2217925,
   "outputTokens": 48771,
   "recordedCostUsd": 2.247999,
   "cost": {
    "Haiku 4.5": 0.7948525,
    "Sonnet 5.5": 1.589705,
    "Opus 5.5": 2.735825,
    "Fable 5.1": 6.28508125,
    "policy (Opus strong, Haiku ancillary)": 2.5498178,
    "split (Sonnet main line, Haiku ancillary)": 1.5296978
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "session_pr348_5",
   "calls": 49,
   "inputTokens": 5726251,
   "cachedTokens": 5328200,
   "outputTokens": 59512,
   "recordedCostUsd": 3.252757,
   "cost": {
    "Haiku 4.5": 1.228431,
    "Sonnet 5.5": 2.456862,
    "Opus 5.5": 3.848084,
    "Fable 5.1": 8.28816,
    "policy (Opus strong, Haiku ancillary)": 3.6922181,
    "split (Sonnet main line, Haiku ancillary)": 2.4066321
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "sphinx_doc_sphinx_10435_1",
   "calls": 63,
   "inputTokens": 6898574,
   "cachedTokens": 6264545,
   "outputTokens": 66774,
   "recordedCostUsd": 4.456506,
   "cost": {
    "Haiku 4.5": 1.5943535,
    "Sonnet 5.5": 3.188707,
    "Opus 5.5": 5.124505,
    "Fable 5.1": 11.24512625,
    "policy (Opus strong, Haiku ancillary)": 4.948155600000001,
    "split (Sonnet main line, Haiku ancillary)": 3.1288156000000003
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "sphinx_doc_sphinx_9698_1",
   "calls": 47,
   "inputTokens": 4776121,
   "cachedTokens": 4469637,
   "outputTokens": 40807,
   "recordedCostUsd": 2.5277369999999997,
   "cost": {
    "Haiku 4.5": 0.9574826999999999,
    "Sonnet 5.5": 1.9149653999999998,
    "Opus 5.5": 2.9360033999999997,
    "Fable 5.1": 6.22259925,
    "policy (Opus strong, Haiku ancillary)": 2.7687837,
    "split (Sonnet main line, Haiku ancillary)": 1.8639377000000001
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "sympy_sympy_18189_1",
   "calls": 13,
   "inputTokens": 996024,
   "cachedTokens": 846144,
   "outputTokens": 12113,
   "recordedCostUsd": 0.8898269999999999,
   "cost": {
    "Haiku 4.5": 0.2950594,
    "Sonnet 5.5": 0.5901188,
    "Opus 5.5": 1.0110088000000002,
    "Fable 5.1": 2.315986,
    "policy (Opus strong, Haiku ancillary)": 0.8950684999999999,
    "split (Sonnet main line, Haiku ancillary)": 0.5511305
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "sympy_sympy_20428_1",
   "calls": 68,
   "inputTokens": 8297349,
   "cachedTokens": 7704037,
   "outputTokens": 96648,
   "recordedCostUsd": 4.8802579999999995,
   "cost": {
    "Haiku 4.5": 1.8469557,
    "Sonnet 5.5": 3.6939114,
    "Opus 5.5": 5.8470154,
    "Fable 5.1": 12.69152925,
    "policy (Opus strong, Haiku ancillary)": 5.7363833,
    "split (Sonnet main line, Haiku ancillary)": 3.6567833000000003
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "sympy_sympy_20428_2",
   "calls": 67,
   "inputTokens": 7486904,
   "cachedTokens": 6999746,
   "outputTokens": 110780,
   "recordedCostUsd": 4.456103,
   "cost": {
    "Haiku 4.5": 1.7410326,
    "Sonnet 5.5": 3.4820652,
    "Opus 5.5": 5.5641812,
    "Fable 5.1": 12.1605165,
    "policy (Opus strong, Haiku ancillary)": 5.4184371,
    "split (Sonnet main line, Haiku ancillary)": 3.4391751000000004
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "sympy_sympy_22456_1",
   "calls": 27,
   "inputTokens": 2662156,
   "cachedTokens": 2484542,
   "outputTokens": 46735,
   "recordedCostUsd": 1.674597,
   "cost": {
    "Haiku 4.5": 0.6597432,
    "Sonnet 5.5": 1.3194864,
    "Opus 5.5": 2.1420643999999998,
    "Fable 5.1": 4.7340255,
    "policy (Opus strong, Haiku ancillary)": 1.9246568000000002,
    "split (Sonnet main line, Haiku ancillary)": 1.2460108
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "uvicorn_pr3036_1",
   "calls": 63,
   "inputTokens": 6506123,
   "cachedTokens": 5898824,
   "outputTokens": 85581,
   "recordedCostUsd": 4.464516,
   "cost": {
    "Haiku 4.5": 1.6250863999999998,
    "Sonnet 5.5": 3.2501727999999996,
    "Opus 5.5": 5.3205808,
    "Fable 5.1": 11.826746,
    "policy (Opus strong, Haiku ancillary)": 4.940384600000001,
    "split (Sonnet main line, Haiku ancillary)": 3.1295826
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "uvicorn_pr3036_2",
   "calls": 64,
   "inputTokens": 7067691,
   "cachedTokens": 6513434,
   "outputTokens": 69388,
   "recordedCostUsd": 4.213336,
   "cost": {
    "Haiku 4.5": 1.5525403999999998,
    "Sonnet 5.5": 3.1050807999999996,
    "Opus 5.5": 4.9074748,
    "Fable 5.1": 10.6403285,
    "policy (Opus strong, Haiku ancillary)": 4.5837924999999995,
    "split (Sonnet main line, Haiku ancillary)": 3.0037904999999996
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "uvicorn_pr3036_4",
   "calls": 70,
   "inputTokens": 7691079,
   "cachedTokens": 7147975,
   "outputTokens": 83085,
   "recordedCostUsd": 4.432582,
   "cost": {
    "Haiku 4.5": 1.6733265,
    "Sonnet 5.5": 3.346653,
    "Opus 5.5": 5.263711,
    "Fable 5.1": 11.37228375,
    "policy (Opus strong, Haiku ancillary)": 4.9813642,
    "split (Sonnet main line, Haiku ancillary)": 3.2598682
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-run",
   "run": "uvicorn_pr3036_5",
   "calls": 59,
   "inputTokens": 6303921,
   "cachedTokens": 5711481,
   "outputTokens": 76036,
   "recordedCostUsd": 4.272171999999999,
   "cost": {
    "Haiku 4.5": 1.5437681,
    "Sonnet 5.5": 3.0875362,
    "Opus 5.5": 5.0327762,
    "Fable 5.1": 11.15407025,
    "policy (Opus strong, Haiku ancillary)": 4.718981,
    "split (Sonnet main line, Haiku ancillary)": 2.9891490000000003
   }
  },
  {
   "batch": "batch-1",
   "type": "economics-totals",
   "runs": 50,
   "calls": 2467,
   "tokenCalls": 2362,
   "inputTokens": 248127848,
   "cachedTokens": 230850011,
   "outputTokens": 2781731,
   "callsOver200k": 2,
   "failedCalls": 3,
   "recordedCostUsd": 153.1114769999999,
   "recordedCostTokenCallsUsd": 143.08895699999997,
   "compactCalls": 105,
   "compactRecordedCostUsd": 10.022519999999998,
   "cacheReadShare": 0.9303671992512506,
   "actual": {
    "provider": [
     "claude-code"
    ],
    "model": [
     "claude-sonnet-5-5"
    ],
    "routing": "off: every recorded call ran on one model"
   },
   "assumptions": [
    "Same tokens, same cache-read share and same number of calls on every model; a different model would take a different path and number of turns.",
    "Cache writes are not recorded; uncached input is priced at the plain input rate (cache writes cost 1.25x), so every figure is a slight underestimate.",
    "Tokenizers differ between model families; token counts are not converted.",
    "runtime.compact calls record a cost but no tokens (105 calls); they are left out of the repricing.",
    "Policy mix: tiers from the platform routing policy (act, research, verify, review on the strong tier; learning, memory and compact on economy; other purposes standard), with the claude-code ladder economy=Haiku, standard=Sonnet, strong=Opus. The ladder alias \"opus\" is priced as Opus 5.5 here.",
    "Haiku 4.5 has a 200K window; 2 recorded calls had more input than that and could not run on it as recorded."
   ],
   "prices": {
    "claude-haiku-4-5": {
     "label": "Haiku 4.5",
     "input": 1,
     "output": 5,
     "cacheRead": 0.1,
     "contextWindow": 200000
    },
    "claude-sonnet-5-5": {
     "label": "Sonnet 5.5",
     "input": 2,
     "output": 10,
     "cacheRead": 0.2,
     "contextWindow": 1000000
    },
    "claude-opus-5-5": {
     "label": "Opus 5.5",
     "input": 4,
     "output": 20,
     "cacheRead": 0.2,
     "contextWindow": 1000000
    },
    "claude-fable-5-1": {
     "label": "Fable 5.1",
     "input": 10,
     "output": 50,
     "cacheRead": 0.25,
     "contextWindow": 1000000
    }
   }
  }
 ],
 "protocolNote": "The run protocol was declared before the first run. Its public summary is the Method section of the study page."
}