{
  "schema": "thruline-bench-release/1",
  "benchmark": "decision-capture",
  "version": "v0.1",
  "updated": "2026-10-31",
  "set": {
    "name": "pilot-v2",
    "sha256": "dce8095600b3ca23f06063dc801ede7d9ab6eeec997764098bc5d8e9cab78e89",
    "scenarios": 16,
    "public": 16,
    "heldOut": 0,
    "types": [
      "made",
      "reversed",
      "someone else decided",
      "partial plan",
      "deferred",
      "delegated",
      "not a decision",
      "off the record"
    ],
    "phrasings": [
      "blunt",
      "multi-turn"
    ]
  },
  "runs": [
    {
      "id": "2026-10-31-00-51",
      "date": "2026-10-31",
      "sessions": 224,
      "spend": 5.2354645481,
      "scoredAt": "2026-10-31T17:23:27.431Z",
      "scoredWith": {
        "set": "pilot-v2",
        "setSha": "dce8095600b3ca23f06063dc801ede7d9ab6eeec997764098bc5d8e9cab78e89",
        "scorer": "04087fe3801f",
        "postHoc": true
      }
    },
    {
      "id": "2026-10-31-02-16",
      "date": "2026-10-31",
      "sessions": 120,
      "spend": 3.895828812700001,
      "scoredAt": "2026-10-31T17:23:27.608Z",
      "commit": "7734efc012ad059619aa98a139474d622aecbd98",
      "scoredWith": {
        "set": "pilot-v2",
        "setSha": "dce8095600b3ca23f06063dc801ede7d9ab6eeec997764098bc5d8e9cab78e89",
        "scorer": "04087fe3801f",
        "postHoc": true
      }
    }
  ],
  "default": {
    "tools": "few",
    "prompt": "generic"
  },
  "headline": {
    "run": "2026-10-31-00-51",
    "measure": "right",
    "off": {
      "k": 52,
      "n": 63,
      "rate": 0.8253968253968254,
      "ci": [
        0.713783402655744,
        0.8996070607682313
      ]
    },
    "on": {
      "k": 71,
      "n": 74,
      "rate": 0.9594594594594594,
      "ci": [
        0.8874510681944234,
        0.9861178204204044
      ]
    },
    "onlyWith": 9,
    "onlyWithout": 1,
    "p": 0.021484375,
    "strict": {
      "off": {
        "k": 45,
        "n": 63,
        "rate": 0.7142857142857143,
        "ci": [
          0.5929704958526203,
          0.8109695654265832
        ]
      },
      "on": {
        "k": 66,
        "n": 74,
        "rate": 0.8918918918918919,
        "ci": [
          0.8009095372797369,
          0.9441933383034987
        ]
      },
      "p": 0.00634765625
    }
  },
  "models": [
    {
      "id": "anthropic/claude-sonnet-5.5",
      "name": "Claude Sonnet 5.5",
      "provider": "Anthropic",
      "added": "2026-10-31",
      "rows": [
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 16,
          "failed": 0,
          "right": {
            "k": 14,
            "n": 16,
            "rate": 0.875,
            "ci": [
              0.6397666856325173,
              0.9650232506629428
            ]
          },
          "strict": {
            "k": 13,
            "n": 16,
            "rate": 0.8125,
            "ci": [
              0.5699064013154177,
              0.9340852122641324
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 4,
            "rate": 0,
            "ci": [
              0,
              0.48990002040399916
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 2,
            "rate": 0,
            "ci": [
              0,
              0.6576280471103807
            ]
          },
          "invented": {
            "k": 0,
            "n": 17,
            "rate": 0,
            "ci": [
              0,
              0.18432366037156456
            ]
          },
          "duplicates": {
            "k": 5,
            "n": 8,
            "rate": 0.625,
            "ci": [
              0.30573785458380187,
              0.863158240538479
            ]
          },
          "points": {
            "mean": 3.823529411764706,
            "ci": [
              2.933333333333333,
              4.529411764705882
            ],
            "reports": 17,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 15
              },
              "goal": {
                "mean": 1,
                "n": 9
              },
              "metric": {
                "mean": 1,
                "n": 6
              },
              "direction": {
                "mean": 1,
                "n": 6
              },
              "why": {
                "mean": 1,
                "n": 7
              },
              "judge_by": {
                "mean": 1,
                "n": 3
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 5080,
            "perSession": {
              "p25": 30480,
              "median": 35711,
              "p75": 46013
            },
            "sessions": 16
          },
          "costPerSession": 0.109588625,
          "costPerCapture": 0.21917725
        },
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 16,
          "failed": 0,
          "right": {
            "k": 13,
            "n": 16,
            "rate": 0.8125,
            "ci": [
              0.5699064013154177,
              0.9340852122641324
            ]
          },
          "strict": {
            "k": 12,
            "n": 16,
            "rate": 0.75,
            "ci": [
              0.5050123107060135,
              0.8981809801576265
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 1,
            "n": 4,
            "rate": 0.25,
            "ci": [
              0.045586062644636216,
              0.6993639475573634
            ]
          },
          "offRecord": {
            "k": 1,
            "n": 2,
            "rate": 0.5,
            "ci": [
              0.09452865480086614,
              0.9054713451991339
            ]
          },
          "invented": {
            "k": 0,
            "n": 20,
            "rate": 0,
            "ci": [
              0,
              0.16113012549493322
            ]
          },
          "duplicates": {
            "k": 6,
            "n": 8,
            "rate": 0.75,
            "ci": [
              0.40926987910258916,
              0.9285223111419724
            ]
          },
          "points": {
            "mean": 4,
            "ci": [
              3.2941176470588234,
              4.62962962962963
            ],
            "reports": 20,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 18
              },
              "why": {
                "mean": 0.8181818181818182,
                "n": 11
              },
              "goal": {
                "mean": 1,
                "n": 10
              },
              "direction": {
                "mean": 1,
                "n": 10
              },
              "metric": {
                "mean": 1,
                "n": 7
              },
              "judge_by": {
                "mean": 1,
                "n": 4
              },
              "decided_by": {
                "mean": 1,
                "n": 4
              }
            }
          },
          "context": {
            "perRequest": 2058,
            "perSession": {
              "p25": 14679,
              "median": 16598,
              "p75": 21135
            },
            "sessions": 16
          },
          "costPerSession": 0.10480987500000001,
          "costPerCapture": 0.20961975000000002
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "compact",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 7,
            "n": 8,
            "rate": 0.875,
            "ci": [
              0.5291051942301386,
              0.9775830911367038
            ]
          },
          "strict": {
            "k": 7,
            "n": 8,
            "rate": 0.875,
            "ci": [
              0.5291051942301386,
              0.9775830911367038
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 9,
            "rate": 0,
            "ci": [
              0,
              0.2991527535509594
            ]
          },
          "duplicates": {
            "k": 1,
            "n": 8,
            "rate": 0.125,
            "ci": [
              0.02241690886329617,
              0.4708948057698615
            ]
          },
          "points": {
            "mean": 5.111111111111111,
            "ci": [
              4.375,
              5.9
            ],
            "reports": 9,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 9
              },
              "why": {
                "mean": 1,
                "n": 5
              },
              "goal": {
                "mean": 1,
                "n": 7
              },
              "metric": {
                "mean": 1,
                "n": 5
              },
              "direction": {
                "mean": 1,
                "n": 7
              },
              "judge_by": {
                "mean": 1,
                "n": 2
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 3063,
            "perSession": {
              "p25": 24778,
              "median": 31205,
              "p75": 34096
            },
            "sessions": 8
          },
          "costPerSession": 0.11166849999999999,
          "costPerCapture": 0.11166849999999999
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "strict": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "duplicates": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "points": {
            "mean": 5.625,
            "ci": [
              4.75,
              6.5
            ],
            "reports": 8,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 8
              },
              "goal": {
                "mean": 1,
                "n": 6
              },
              "metric": {
                "mean": 1,
                "n": 6
              },
              "direction": {
                "mean": 1,
                "n": 6
              },
              "judge_by": {
                "mean": 1,
                "n": 4
              },
              "why": {
                "mean": 1,
                "n": 5
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 5291,
            "perSession": {
              "p25": 42602,
              "median": 53219,
              "p75": 58599
            },
            "sessions": 8
          },
          "costPerSession": 0.13539549999999997,
          "costPerCapture": 0.13539549999999997
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 6,
            "n": 8,
            "rate": 0.75,
            "ci": [
              0.40926987910258916,
              0.9285223111419724
            ]
          },
          "strict": {
            "k": 6,
            "n": 8,
            "rate": 0.75,
            "ci": [
              0.40926987910258916,
              0.9285223111419724
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "duplicates": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "points": {
            "mean": 4.75,
            "ci": [
              3.375,
              6
            ],
            "reports": 8,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 8
              },
              "why": {
                "mean": 0.42857142857142855,
                "n": 7
              },
              "goal": {
                "mean": 1,
                "n": 5
              },
              "direction": {
                "mean": 1,
                "n": 5
              },
              "judge_by": {
                "mean": 1,
                "n": 4
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              },
              "metric": {
                "mean": 1,
                "n": 4
              }
            }
          },
          "context": {
            "perRequest": 2238,
            "perSession": {
              "p25": 20548,
              "median": 24904,
              "p75": 27509
            },
            "sessions": 8
          },
          "costPerSession": 0.1237095,
          "costPerCapture": 0.1237095
        }
      ]
    },
    {
      "id": "anthropic/claude-haiku-5.5",
      "name": "Claude Haiku 5.5",
      "provider": "Anthropic",
      "added": "2026-10-31",
      "rows": [
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 16,
          "failed": 0,
          "right": {
            "k": 13,
            "n": 16,
            "rate": 0.8125,
            "ci": [
              0.5699064013154177,
              0.9340852122641324
            ]
          },
          "strict": {
            "k": 12,
            "n": 16,
            "rate": 0.75,
            "ci": [
              0.5050123107060135,
              0.8981809801576265
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 1,
            "n": 4,
            "rate": 0.25,
            "ci": [
              0.045586062644636216,
              0.6993639475573634
            ]
          },
          "offRecord": {
            "k": 1,
            "n": 2,
            "rate": 0.5,
            "ci": [
              0.09452865480086614,
              0.9054713451991339
            ]
          },
          "invented": {
            "k": 0,
            "n": 22,
            "rate": 0,
            "ci": [
              1.3877787807814457e-17,
              0.14865952572596125
            ]
          },
          "duplicates": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "points": {
            "mean": 3.272727272727273,
            "ci": [
              2.388888888888889,
              3.8461538461538463
            ],
            "reports": 22,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 20
              },
              "why": {
                "mean": 0.5,
                "n": 12
              },
              "goal": {
                "mean": 0.8181818181818182,
                "n": 11
              },
              "direction": {
                "mean": 1,
                "n": 10
              },
              "metric": {
                "mean": 1,
                "n": 7
              },
              "judge_by": {
                "mean": 1,
                "n": 4
              },
              "decided_by": {
                "mean": 1,
                "n": 4
              }
            }
          },
          "context": {
            "perRequest": 2058,
            "perSession": {
              "p25": 14406,
              "median": 18731,
              "p75": 22697
            },
            "sessions": 16
          },
          "costPerSession": 0.00503810625,
          "costPerCapture": 0.0100762125
        },
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 16,
          "failed": 0,
          "right": {
            "k": 16,
            "n": 16,
            "rate": 1,
            "ci": [
              0.8063865817272802,
              1
            ]
          },
          "strict": {
            "k": 15,
            "n": 16,
            "rate": 0.9375,
            "ci": [
              0.7167073163197013,
              0.9888809426916687
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 4,
            "rate": 0,
            "ci": [
              0,
              0.48990002040399916
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 2,
            "rate": 0,
            "ci": [
              0,
              0.6576280471103807
            ]
          },
          "invented": {
            "k": 0,
            "n": 12,
            "rate": 0,
            "ci": [
              0,
              0.2425007574992425
            ]
          },
          "duplicates": {
            "k": 3,
            "n": 8,
            "rate": 0.375,
            "ci": [
              0.13684175946152105,
              0.6942621454161981
            ]
          },
          "points": {
            "mean": 3.4166666666666665,
            "ci": [
              2.75,
              3.9166666666666665
            ],
            "reports": 12,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 11
              },
              "why": {
                "mean": 0.6666666666666666,
                "n": 6
              },
              "goal": {
                "mean": 1,
                "n": 5
              },
              "metric": {
                "mean": 1,
                "n": 3
              },
              "direction": {
                "mean": 1,
                "n": 4
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 5080,
            "perSession": {
              "p25": 30584,
              "median": 35576,
              "p75": 40858
            },
            "sessions": 16
          },
          "costPerSession": 0.0052918375,
          "costPerCapture": 0.010583675
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 6,
            "n": 8,
            "rate": 0.75,
            "ci": [
              0.40926987910258916,
              0.9285223111419724
            ]
          },
          "strict": {
            "k": 6,
            "n": 8,
            "rate": 0.75,
            "ci": [
              0.40926987910258916,
              0.9285223111419724
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 9,
            "rate": 0,
            "ci": [
              0,
              0.2991527535509594
            ]
          },
          "duplicates": {
            "k": 1,
            "n": 8,
            "rate": 0.125,
            "ci": [
              0.02241690886329617,
              0.4708948057698615
            ]
          },
          "points": {
            "mean": 4.777777777777778,
            "ci": [
              4,
              5.5
            ],
            "reports": 9,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 9
              },
              "goal": {
                "mean": 1,
                "n": 6
              },
              "metric": {
                "mean": 1,
                "n": 5
              },
              "direction": {
                "mean": 1,
                "n": 6
              },
              "judge_by": {
                "mean": 1,
                "n": 4
              },
              "why": {
                "mean": 0.6,
                "n": 5
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 2238,
            "perSession": {
              "p25": 20643,
              "median": 22770,
              "p75": 25061
            },
            "sessions": 8
          },
          "costPerSession": 0.0056075249999999995,
          "costPerCapture": 0.0056075249999999995
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "compact",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 7,
            "n": 8,
            "rate": 0.875,
            "ci": [
              0.5291051942301386,
              0.9775830911367038
            ]
          },
          "strict": {
            "k": 7,
            "n": 8,
            "rate": 0.875,
            "ci": [
              0.5291051942301386,
              0.9775830911367038
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "duplicates": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "points": {
            "mean": 5,
            "ci": [
              4.375,
              5.75
            ],
            "reports": 8,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 8
              },
              "why": {
                "mean": 1,
                "n": 5
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              },
              "goal": {
                "mean": 1,
                "n": 5
              },
              "direction": {
                "mean": 1,
                "n": 5
              },
              "metric": {
                "mean": 1,
                "n": 4
              },
              "judge_by": {
                "mean": 1,
                "n": 3
              }
            }
          },
          "context": {
            "perRequest": 3063,
            "perSession": {
              "p25": 24855,
              "median": 27957,
              "p75": 31237
            },
            "sessions": 8
          },
          "costPerSession": 0.005491725,
          "costPerCapture": 0.005491725
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 7,
            "n": 8,
            "rate": 0.875,
            "ci": [
              0.5291051942301386,
              0.9775830911367038
            ]
          },
          "strict": {
            "k": 7,
            "n": 8,
            "rate": 0.875,
            "ci": [
              0.5291051942301386,
              0.9775830911367038
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 9,
            "rate": 0,
            "ci": [
              0,
              0.2991527535509594
            ]
          },
          "duplicates": {
            "k": 1,
            "n": 8,
            "rate": 0.125,
            "ci": [
              0.02241690886329617,
              0.4708948057698615
            ]
          },
          "points": {
            "mean": 5,
            "ci": [
              3.6666666666666665,
              6.222222222222222
            ],
            "reports": 9,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 9
              },
              "goal": {
                "mean": 1,
                "n": 7
              },
              "metric": {
                "mean": 0.7142857142857143,
                "n": 7
              },
              "direction": {
                "mean": 0.7142857142857143,
                "n": 7
              },
              "why": {
                "mean": 0.6666666666666666,
                "n": 6
              },
              "judge_by": {
                "mean": 1,
                "n": 4
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 5291,
            "perSession": {
              "p25": 37301,
              "median": 48045,
              "p75": 53383
            },
            "sessions": 8
          },
          "costPerSession": 0.0067306625,
          "costPerCapture": 0.0067306625
        }
      ]
    },
    {
      "id": "google/gemini-3.8-flash",
      "name": "Gemini 3.8 Flash",
      "provider": "Google",
      "added": "2026-10-31",
      "rows": [
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 16,
          "failed": 5,
          "right": {
            "k": 11,
            "n": 11,
            "rate": 1,
            "ci": [
              0.7411599827511859,
              1
            ]
          },
          "strict": {
            "k": 10,
            "n": 11,
            "rate": 0.9090909090909091,
            "ci": [
              0.6226353745137962,
              0.9837682477371741
            ]
          },
          "capture": {
            "k": 7,
            "n": 7,
            "rate": 1,
            "ci": [
              0.6456611570247934,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 2,
            "rate": 0,
            "ci": [
              0,
              0.6576280471103807
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 11,
            "rate": 0,
            "ci": [
              2.7755575615628914e-17,
              0.2588400172488141
            ]
          },
          "duplicates": {
            "k": 3,
            "n": 7,
            "rate": 0.42857142857142855,
            "ci": [
              0.15821692226262685,
              0.7495457695909742
            ]
          },
          "points": {
            "mean": 3.5454545454545454,
            "ci": [
              2.8,
              4.3
            ],
            "reports": 11,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 10
              },
              "goal": {
                "mean": 0.5,
                "n": 4
              },
              "metric": {
                "mean": 1,
                "n": 3
              },
              "direction": {
                "mean": 1,
                "n": 3
              },
              "decided_by": {
                "mean": 1,
                "n": 3
              },
              "why": {
                "mean": 1,
                "n": 6
              },
              "judge_by": {
                "mean": 1,
                "n": 1
              }
            }
          },
          "context": {
            "perRequest": 5080,
            "perSession": {
              "p25": 30601,
              "median": 45964,
              "p75": 61021
            },
            "sessions": 11
          },
          "costPerSession": 0.032709571875,
          "costPerCapture": 0.07476473571428573
        },
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 16,
          "failed": 6,
          "right": {
            "k": 8,
            "n": 10,
            "rate": 0.8,
            "ci": [
              0.49015684672072335,
              0.9433190520193067
            ]
          },
          "strict": {
            "k": 6,
            "n": 10,
            "rate": 0.6,
            "ci": [
              0.3126695474501864,
              0.8318224187964902
            ]
          },
          "capture": {
            "k": 7,
            "n": 7,
            "rate": 1,
            "ci": [
              0.6456611570247934,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 1,
            "rate": 0,
            "ci": [
              0,
              0.7934567085261071
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 20,
            "rate": 0,
            "ci": [
              0,
              0.16113012549493322
            ]
          },
          "duplicates": {
            "k": 6,
            "n": 7,
            "rate": 0.8571428571428571,
            "ci": [
              0.48686549668097007,
              0.9743210440510253
            ]
          },
          "points": {
            "mean": 3.55,
            "ci": [
              2.2666666666666666,
              4.291666666666667
            ],
            "reports": 20,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 18
              },
              "goal": {
                "mean": 0.8333333333333334,
                "n": 12
              },
              "direction": {
                "mean": 1,
                "n": 10
              },
              "metric": {
                "mean": 1,
                "n": 8
              },
              "judge_by": {
                "mean": 0.5,
                "n": 4
              },
              "why": {
                "mean": 0.07692307692307693,
                "n": 13
              },
              "decided_by": {
                "mean": 1,
                "n": 3
              }
            }
          },
          "context": {
            "perRequest": 2058,
            "perSession": {
              "p25": 16872,
              "median": 24935,
              "p75": 27261
            },
            "sessions": 10
          },
          "costPerSession": 0.0201969140625,
          "costPerCapture": 0.046164375
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 8,
          "failed": 1,
          "right": {
            "k": 5,
            "n": 7,
            "rate": 0.7142857142857143,
            "ci": [
              0.35892909014821267,
              0.9177828342909844
            ]
          },
          "strict": {
            "k": 5,
            "n": 7,
            "rate": 0.7142857142857143,
            "ci": [
              0.35892909014821267,
              0.9177828342909844
            ]
          },
          "capture": {
            "k": 7,
            "n": 7,
            "rate": 1,
            "ci": [
              0.6456611570247934,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "duplicates": {
            "k": 1,
            "n": 7,
            "rate": 0.14285714285714285,
            "ci": [
              0.02567895594897479,
              0.51313450331903
            ]
          },
          "points": {
            "mean": 5,
            "ci": [
              4,
              5.857142857142857
            ],
            "reports": 8,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 8
              },
              "why": {
                "mean": 0.6666666666666666,
                "n": 6
              },
              "goal": {
                "mean": 1,
                "n": 6
              },
              "direction": {
                "mean": 1,
                "n": 6
              },
              "judge_by": {
                "mean": 1,
                "n": 3
              },
              "metric": {
                "mean": 1,
                "n": 4
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 2238,
            "perSession": {
              "p25": 20400,
              "median": 29514,
              "p75": 38512
            },
            "sessions": 7
          },
          "costPerSession": 0.026209425000000005,
          "costPerCapture": 0.029953628571428578
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 8,
          "failed": 2,
          "right": {
            "k": 6,
            "n": 6,
            "rate": 1,
            "ci": [
              0.6096569663469354,
              0.9999999999999999
            ]
          },
          "strict": {
            "k": 6,
            "n": 6,
            "rate": 1,
            "ci": [
              0.6096569663469354,
              0.9999999999999999
            ]
          },
          "capture": {
            "k": 6,
            "n": 6,
            "rate": 1,
            "ci": [
              0.6096569663469354,
              0.9999999999999999
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 6,
            "rate": 0,
            "ci": [
              0,
              0.3903430336530645
            ]
          },
          "duplicates": {
            "k": 0,
            "n": 6,
            "rate": 0,
            "ci": [
              0,
              0.3903430336530645
            ]
          },
          "points": {
            "mean": 4,
            "ci": [
              2.6666666666666665,
              5.333333333333333
            ],
            "reports": 6,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 6
              },
              "goal": {
                "mean": 1,
                "n": 3
              },
              "metric": {
                "mean": 0.3333333333333333,
                "n": 3
              },
              "direction": {
                "mean": 0.3333333333333333,
                "n": 3
              },
              "why": {
                "mean": 1,
                "n": 4
              },
              "judge_by": {
                "mean": 1,
                "n": 1
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 5291,
            "perSession": {
              "p25": 26648,
              "median": 47883,
              "p75": 53199
            },
            "sessions": 6
          },
          "costPerSession": 0.029327090624999996,
          "costPerCapture": 0.03910278749999999
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "compact",
          "split": "all",
          "sessions": 8,
          "failed": 1,
          "right": {
            "k": 6,
            "n": 7,
            "rate": 0.8571428571428571,
            "ci": [
              0.48686549668097007,
              0.9743210440510253
            ]
          },
          "strict": {
            "k": 6,
            "n": 7,
            "rate": 0.8571428571428571,
            "ci": [
              0.48686549668097007,
              0.9743210440510253
            ]
          },
          "capture": {
            "k": 7,
            "n": 7,
            "rate": 1,
            "ci": [
              0.6456611570247934,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "duplicates": {
            "k": 1,
            "n": 7,
            "rate": 0.14285714285714285,
            "ci": [
              0.02567895594897479,
              0.51313450331903
            ]
          },
          "points": {
            "mean": 4.625,
            "ci": [
              3.142857142857143,
              5.777777777777778
            ],
            "reports": 8,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 8
              },
              "goal": {
                "mean": 1,
                "n": 6
              },
              "metric": {
                "mean": 0.6,
                "n": 5
              },
              "direction": {
                "mean": 0.6666666666666666,
                "n": 6
              },
              "why": {
                "mean": 1,
                "n": 4
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              },
              "judge_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 3063,
            "perSession": {
              "p25": 18636,
              "median": 33957,
              "p75": 46378
            },
            "sessions": 7
          },
          "costPerSession": 0.029624568750000004,
          "costPerCapture": 0.03385665
        }
      ]
    },
    {
      "id": "deepseek/deepseek-v4.1-flash",
      "name": "DeepSeek V4.1 Flash",
      "provider": "DeepSeek",
      "added": "2026-10-31",
      "rows": [
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 16,
          "failed": 1,
          "right": {
            "k": 15,
            "n": 15,
            "rate": 1,
            "ci": [
              0.7961107336956521,
              1
            ]
          },
          "strict": {
            "k": 14,
            "n": 15,
            "rate": 0.9333333333333333,
            "ci": [
              0.7018292244757601,
              0.9881334113938051
            ]
          },
          "capture": {
            "k": 7,
            "n": 7,
            "rate": 1,
            "ci": [
              0.6456611570247934,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 4,
            "rate": 0,
            "ci": [
              0,
              0.48990002040399916
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 2,
            "rate": 0,
            "ci": [
              0,
              0.6576280471103807
            ]
          },
          "invented": {
            "k": 0,
            "n": 13,
            "rate": 0,
            "ci": [
              0,
              0.22810184305529166
            ]
          },
          "duplicates": {
            "k": 3,
            "n": 7,
            "rate": 0.42857142857142855,
            "ci": [
              0.15821692226262685,
              0.7495457695909742
            ]
          },
          "points": {
            "mean": 3.8461538461538463,
            "ci": [
              2.75,
              4.545454545454546
            ],
            "reports": 13,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 12
              },
              "why": {
                "mean": 1,
                "n": 5
              },
              "goal": {
                "mean": 1,
                "n": 7
              },
              "metric": {
                "mean": 1,
                "n": 5
              },
              "direction": {
                "mean": 1,
                "n": 5
              },
              "judge_by": {
                "mean": 1,
                "n": 1
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 5080,
            "perSession": {
              "p25": 25519,
              "median": 35666,
              "p75": 51168
            },
            "sessions": 15
          },
          "costPerSession": 0.0035242611312499995,
          "costPerCapture": 0.008055454014285713
        },
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 16,
          "failed": 8,
          "right": {
            "k": 7,
            "n": 8,
            "rate": 0.875,
            "ci": [
              0.5291051942301386,
              0.9775830911367038
            ]
          },
          "strict": {
            "k": 6,
            "n": 8,
            "rate": 0.75,
            "ci": [
              0.40926987910258916,
              0.9285223111419724
            ]
          },
          "capture": {
            "k": 4,
            "n": 4,
            "rate": 1,
            "ci": [
              0.5100999795960008,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 2,
            "rate": 0,
            "ci": [
              0,
              0.6576280471103807
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 1,
            "rate": 0,
            "ci": [
              0,
              0.7934567085261071
            ]
          },
          "invented": {
            "k": 0,
            "n": 12,
            "rate": 0,
            "ci": [
              0,
              0.2425007574992425
            ]
          },
          "duplicates": {
            "k": 4,
            "n": 4,
            "rate": 1,
            "ci": [
              0.5100999795960008,
              1
            ]
          },
          "points": {
            "mean": 3.1666666666666665,
            "ci": [
              2,
              3.9285714285714284
            ],
            "reports": 12,
            "fields": {
              "why": {
                "mean": 0,
                "n": 8
              },
              "summary": {
                "mean": 1,
                "n": 11
              },
              "goal": {
                "mean": 1,
                "n": 5
              },
              "direction": {
                "mean": 1,
                "n": 5
              },
              "metric": {
                "mean": 1,
                "n": 4
              },
              "judge_by": {
                "mean": 1,
                "n": 1
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 2058,
            "perSession": {
              "p25": 24785,
              "median": 24935,
              "p75": 31527
            },
            "sessions": 8
          },
          "costPerSession": 0.0028996118750000003,
          "costPerCapture": 0.011598447500000001
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 8,
          "failed": 4,
          "right": {
            "k": 3,
            "n": 4,
            "rate": 0.75,
            "ci": [
              0.3006360524426366,
              0.9544139373553637
            ]
          },
          "strict": {
            "k": 3,
            "n": 4,
            "rate": 0.75,
            "ci": [
              0.3006360524426366,
              0.9544139373553637
            ]
          },
          "capture": {
            "k": 4,
            "n": 4,
            "rate": 1,
            "ci": [
              0.5100999795960008,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 1,
            "n": 4,
            "rate": 0.25,
            "ci": [
              0.045586062644636216,
              0.6993639475573634
            ]
          },
          "duplicates": {
            "k": 0,
            "n": 4,
            "rate": 0,
            "ci": [
              0,
              0.48990002040399916
            ]
          },
          "points": {
            "mean": 3.5,
            "ci": [
              1.5,
              6
            ],
            "reports": 4,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 4
              },
              "why": {
                "mean": 0.3333333333333333,
                "n": 3
              },
              "metric": {
                "mean": 1,
                "n": 2
              },
              "direction": {
                "mean": 1,
                "n": 2
              },
              "judge_by": {
                "mean": 0,
                "n": 2
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              },
              "goal": {
                "mean": 1,
                "n": 1
              }
            }
          },
          "context": {
            "perRequest": 2238,
            "perSession": {
              "p25": 29475,
              "median": 36191,
              "p75": 38880
            },
            "sessions": 4
          },
          "costPerSession": 0.0030453826375,
          "costPerCapture": 0.006090765275
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "strict": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "duplicates": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "points": {
            "mean": 5,
            "ci": [
              4.375,
              5.75
            ],
            "reports": 8,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 8
              },
              "goal": {
                "mean": 1,
                "n": 4
              },
              "metric": {
                "mean": 1,
                "n": 5
              },
              "direction": {
                "mean": 1,
                "n": 4
              },
              "judge_by": {
                "mean": 1,
                "n": 4
              },
              "why": {
                "mean": 1,
                "n": 5
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 5291,
            "perSession": {
              "p25": 47923,
              "median": 53345,
              "p75": 74684
            },
            "sessions": 8
          },
          "costPerSession": 0.0035900827,
          "costPerCapture": 0.0035900827
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "compact",
          "split": "all",
          "sessions": 8,
          "failed": 2,
          "right": {
            "k": 4,
            "n": 6,
            "rate": 0.6666666666666666,
            "ci": [
              0.29998832134152864,
              0.9032306674407831
            ]
          },
          "strict": {
            "k": 4,
            "n": 6,
            "rate": 0.6666666666666666,
            "ci": [
              0.29998832134152864,
              0.9032306674407831
            ]
          },
          "capture": {
            "k": 6,
            "n": 6,
            "rate": 1,
            "ci": [
              0.6096569663469354,
              0.9999999999999999
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "duplicates": {
            "k": 2,
            "n": 6,
            "rate": 0.3333333333333333,
            "ci": [
              0.09676933255921683,
              0.7000116786584712
            ]
          },
          "points": {
            "mean": 4.5,
            "ci": [
              3.75,
              5.571428571428571
            ],
            "reports": 8,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 8
              },
              "goal": {
                "mean": 1,
                "n": 4
              },
              "metric": {
                "mean": 1,
                "n": 5
              },
              "direction": {
                "mean": 1,
                "n": 4
              },
              "why": {
                "mean": 1,
                "n": 4
              },
              "judge_by": {
                "mean": 1,
                "n": 1
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 3063,
            "perSession": {
              "p25": 21891,
              "median": 34012,
              "p75": 46373
            },
            "sessions": 6
          },
          "costPerSession": 0.002946055,
          "costPerCapture": 0.003928073333333333
        }
      ]
    },
    {
      "id": "openai/gpt-6-luna",
      "name": "GPT-6 Luna",
      "provider": "OpenAI",
      "added": "2026-10-31",
      "rows": [
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 16,
          "failed": 3,
          "right": {
            "k": 11,
            "n": 13,
            "rate": 0.8461538461538461,
            "ci": [
              0.5776482949136899,
              0.9567427368172621
            ]
          },
          "strict": {
            "k": 9,
            "n": 13,
            "rate": 0.6923076923076923,
            "ci": [
              0.4236888781333764,
              0.8731950283838191
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 2,
            "rate": 0,
            "ci": [
              0,
              0.6576280471103807
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 1,
            "rate": 0,
            "ci": [
              0,
              0.7934567085261071
            ]
          },
          "invented": {
            "k": 0,
            "n": 25,
            "rate": 0,
            "ci": [
              0,
              0.13319649395317873
            ]
          },
          "duplicates": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "points": {
            "mean": 2.72,
            "ci": [
              1.9,
              3.3548387096774195
            ],
            "reports": 25,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 21
              },
              "decided_by": {
                "mean": 1,
                "n": 4
              },
              "why": {
                "mean": 0,
                "n": 16
              },
              "goal": {
                "mean": 0.2727272727272727,
                "n": 11
              },
              "metric": {
                "mean": 0.75,
                "n": 8
              },
              "direction": {
                "mean": 0.75,
                "n": 8
              },
              "judge_by": {
                "mean": 1,
                "n": 3
              }
            }
          },
          "context": {
            "perRequest": 2058,
            "perSession": {
              "p25": 14597,
              "median": 16595,
              "p75": 23004
            },
            "sessions": 13
          },
          "costPerSession": 0.0010083640625,
          "costPerCapture": 0.002016728125
        },
        {
          "run": "2026-10-31-00-51",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 16,
          "failed": 0,
          "right": {
            "k": 15,
            "n": 16,
            "rate": 0.9375,
            "ci": [
              0.7167073163197013,
              0.9888809426916687
            ]
          },
          "strict": {
            "k": 14,
            "n": 16,
            "rate": 0.875,
            "ci": [
              0.6397666856325173,
              0.9650232506629428
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 4,
            "rate": 0,
            "ci": [
              0,
              0.48990002040399916
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 2,
            "rate": 0,
            "ci": [
              0,
              0.6576280471103807
            ]
          },
          "invented": {
            "k": 0,
            "n": 15,
            "rate": 0,
            "ci": [
              0,
              0.20388926630434784
            ]
          },
          "duplicates": {
            "k": 4,
            "n": 8,
            "rate": 0.5,
            "ci": [
              0.21521252682444186,
              0.7847874731755582
            ]
          },
          "points": {
            "mean": 3.4,
            "ci": [
              2,
              4.4375
            ],
            "reports": 15,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 13
              },
              "why": {
                "mean": 0.5,
                "n": 8
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              },
              "goal": {
                "mean": 1,
                "n": 8
              },
              "metric": {
                "mean": 0.6,
                "n": 5
              },
              "direction": {
                "mean": 0.6,
                "n": 5
              },
              "judge_by": {
                "mean": 1,
                "n": 3
              }
            }
          },
          "context": {
            "perRequest": 5080,
            "perSession": {
              "p25": 25524,
              "median": 40759,
              "p75": 46090
            },
            "sessions": 16
          },
          "costPerSession": 0.0012656924999999996,
          "costPerCapture": 0.0025313849999999993
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "full",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "strict": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "duplicates": {
            "k": 0,
            "n": 8,
            "rate": 0,
            "ci": [
              0,
              0.3244156195108769
            ]
          },
          "points": {
            "mean": 3.625,
            "ci": [
              2.5,
              4.875
            ],
            "reports": 8,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 8
              },
              "goal": {
                "mean": 0.3333333333333333,
                "n": 6
              },
              "metric": {
                "mean": 0.5,
                "n": 4
              },
              "why": {
                "mean": 1,
                "n": 4
              },
              "direction": {
                "mean": 0.3333333333333333,
                "n": 3
              },
              "judge_by": {
                "mean": 1,
                "n": 2
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 5291,
            "perSession": {
              "p25": 31925,
              "median": 42602,
              "p75": 53236
            },
            "sessions": 8
          },
          "costPerSession": 0.0012753618750000002,
          "costPerCapture": 0.0012753618750000002
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "compact",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 7,
            "n": 8,
            "rate": 0.875,
            "ci": [
              0.5291051942301386,
              0.9775830911367038
            ]
          },
          "strict": {
            "k": 7,
            "n": 8,
            "rate": 0.875,
            "ci": [
              0.5291051942301386,
              0.9775830911367038
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 9,
            "rate": 0,
            "ci": [
              0,
              0.2991527535509594
            ]
          },
          "duplicates": {
            "k": 1,
            "n": 8,
            "rate": 0.125,
            "ci": [
              0.02241690886329617,
              0.4708948057698615
            ]
          },
          "points": {
            "mean": 3.5555555555555554,
            "ci": [
              2.9,
              4.375
            ],
            "reports": 9,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 9
              },
              "why": {
                "mean": 0.6,
                "n": 5
              },
              "judge_by": {
                "mean": 1,
                "n": 3
              },
              "goal": {
                "mean": 1,
                "n": 4
              },
              "direction": {
                "mean": 1,
                "n": 2
              },
              "metric": {
                "mean": 1,
                "n": 1
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 3063,
            "perSession": {
              "p25": 18557,
              "median": 21874,
              "p75": 40424
            },
            "sessions": 8
          },
          "costPerSession": 0.001238416875,
          "costPerCapture": 0.001238416875
        },
        {
          "run": "2026-10-31-02-16",
          "tools": "few",
          "prompt": "generic",
          "skill": "none",
          "split": "all",
          "sessions": 8,
          "failed": 0,
          "right": {
            "k": 6,
            "n": 8,
            "rate": 0.75,
            "ci": [
              0.40926987910258916,
              0.9285223111419724
            ]
          },
          "strict": {
            "k": 6,
            "n": 8,
            "rate": 0.75,
            "ci": [
              0.40926987910258916,
              0.9285223111419724
            ]
          },
          "capture": {
            "k": 8,
            "n": 8,
            "rate": 1,
            "ci": [
              0.6755843804891231,
              1
            ]
          },
          "falseReports": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "offRecord": {
            "k": 0,
            "n": 0,
            "rate": 0,
            "ci": [
              0,
              1
            ]
          },
          "invented": {
            "k": 0,
            "n": 10,
            "rate": 0,
            "ci": [
              0,
              0.2775401687666166
            ]
          },
          "duplicates": {
            "k": 1,
            "n": 8,
            "rate": 0.125,
            "ci": [
              0.02241690886329617,
              0.4708948057698615
            ]
          },
          "points": {
            "mean": 3.1,
            "ci": [
              2.0833333333333335,
              4.625
            ],
            "reports": 10,
            "fields": {
              "summary": {
                "mean": 1,
                "n": 9
              },
              "why": {
                "mean": -0.14285714285714285,
                "n": 7
              },
              "decided_by": {
                "mean": 1,
                "n": 2
              },
              "metric": {
                "mean": 0.6,
                "n": 5
              },
              "direction": {
                "mean": 0.6,
                "n": 5
              },
              "goal": {
                "mean": 0.6,
                "n": 5
              },
              "judge_by": {
                "mean": 1,
                "n": 2
              }
            }
          },
          "context": {
            "perRequest": 2238,
            "perSession": {
              "p25": 20425,
              "median": 20759,
              "p75": 29497
            },
            "sessions": 8
          },
          "costPerSession": 0.001118805625,
          "costPerCapture": 0.001118805625
        }
      ]
    }
  ],
  "otherSettings": [
    {
      "model": "anthropic/claude-haiku-5.5",
      "run": "2026-10-31-00-51",
      "tools": "many",
      "prompt": "product",
      "skill": "none",
      "split": "all",
      "sessions": 16,
      "failed": 0,
      "right": {
        "k": 13,
        "n": 16,
        "rate": 0.8125,
        "ci": [
          0.5699064013154177,
          0.9340852122641324
        ]
      },
      "strict": {
        "k": 12,
        "n": 16,
        "rate": 0.75,
        "ci": [
          0.5050123107060135,
          0.8981809801576265
        ]
      },
      "capture": {
        "k": 8,
        "n": 8,
        "rate": 1,
        "ci": [
          0.6755843804891231,
          1
        ]
      },
      "falseReports": {
        "k": 1,
        "n": 4,
        "rate": 0.25,
        "ci": [
          0.045586062644636216,
          0.6993639475573634
        ]
      },
      "offRecord": {
        "k": 1,
        "n": 2,
        "rate": 0.5,
        "ci": [
          0.09452865480086614,
          0.9054713451991339
        ]
      },
      "invented": {
        "k": 0,
        "n": 21,
        "rate": 0,
        "ci": [
          0,
          0.15464382326420195
        ]
      },
      "duplicates": {
        "k": 8,
        "n": 8,
        "rate": 1,
        "ci": [
          0.6755843804891231,
          1
        ]
      },
      "points": {
        "mean": 3.1904761904761907,
        "ci": [
          2.5,
          3.72
        ],
        "reports": 21,
        "fields": {
          "summary": {
            "mean": 1,
            "n": 19
          },
          "decided_by": {
            "mean": 1,
            "n": 4
          },
          "why": {
            "mean": 1,
            "n": 5
          },
          "goal": {
            "mean": 1,
            "n": 8
          },
          "metric": {
            "mean": 1,
            "n": 7
          },
          "direction": {
            "mean": 1,
            "n": 8
          },
          "judge_by": {
            "mean": 1,
            "n": 3
          }
        }
      },
      "context": {
        "perRequest": 2058,
        "perSession": {
          "p25": 14406,
          "median": 16758,
          "p75": 22872
        },
        "sessions": 16
      },
      "costPerSession": 0.016784393749999998,
      "costPerCapture": 0.033568787499999996
    },
    {
      "model": "anthropic/claude-haiku-5.5",
      "run": "2026-10-31-00-51",
      "tools": "many",
      "prompt": "product",
      "skill": "full",
      "split": "all",
      "sessions": 16,
      "failed": 0,
      "right": {
        "k": 16,
        "n": 16,
        "rate": 1,
        "ci": [
          0.8063865817272802,
          1
        ]
      },
      "strict": {
        "k": 15,
        "n": 16,
        "rate": 0.9375,
        "ci": [
          0.7167073163197013,
          0.9888809426916687
        ]
      },
      "capture": {
        "k": 8,
        "n": 8,
        "rate": 1,
        "ci": [
          0.6755843804891231,
          1
        ]
      },
      "falseReports": {
        "k": 0,
        "n": 4,
        "rate": 0,
        "ci": [
          0,
          0.48990002040399916
        ]
      },
      "offRecord": {
        "k": 0,
        "n": 2,
        "rate": 0,
        "ci": [
          0,
          0.6576280471103807
        ]
      },
      "invented": {
        "k": 0,
        "n": 13,
        "rate": 0,
        "ci": [
          0,
          0.22810184305529166
        ]
      },
      "duplicates": {
        "k": 4,
        "n": 8,
        "rate": 0.5,
        "ci": [
          0.21521252682444186,
          0.7847874731755582
        ]
      },
      "points": {
        "mean": 4.076923076923077,
        "ci": [
          3.1818181818181817,
          4.857142857142857
        ],
        "reports": 13,
        "fields": {
          "summary": {
            "mean": 1,
            "n": 12
          },
          "decided_by": {
            "mean": 1,
            "n": 2
          },
          "why": {
            "mean": 1,
            "n": 6
          },
          "goal": {
            "mean": 1,
            "n": 7
          },
          "metric": {
            "mean": 1,
            "n": 5
          },
          "direction": {
            "mean": 1,
            "n": 6
          },
          "judge_by": {
            "mean": 1,
            "n": 2
          }
        }
      },
      "context": {
        "perRequest": 5080,
        "perSession": {
          "p25": 35560,
          "median": 40744,
          "p75": 45993
        },
        "sessions": 16
      },
      "costPerSession": 0.01708014375,
      "costPerCapture": 0.0341602875
    },
    {
      "model": "anthropic/claude-haiku-5.5",
      "run": "2026-10-31-00-51",
      "tools": "none",
      "prompt": "none",
      "skill": "none",
      "split": "all",
      "sessions": 16,
      "failed": 0,
      "right": {
        "k": 13,
        "n": 16,
        "rate": 0.8125,
        "ci": [
          0.5699064013154177,
          0.9340852122641324
        ]
      },
      "strict": {
        "k": 12,
        "n": 16,
        "rate": 0.75,
        "ci": [
          0.5050123107060135,
          0.8981809801576265
        ]
      },
      "capture": {
        "k": 8,
        "n": 8,
        "rate": 1,
        "ci": [
          0.6755843804891231,
          1
        ]
      },
      "falseReports": {
        "k": 1,
        "n": 4,
        "rate": 0.25,
        "ci": [
          0.045586062644636216,
          0.6993639475573634
        ]
      },
      "offRecord": {
        "k": 1,
        "n": 2,
        "rate": 0.5,
        "ci": [
          0.09452865480086614,
          0.9054713451991339
        ]
      },
      "invented": {
        "k": 0,
        "n": 23,
        "rate": 0,
        "ci": [
          0,
          0.14312112541726274
        ]
      },
      "duplicates": {
        "k": 8,
        "n": 8,
        "rate": 1,
        "ci": [
          0.6755843804891231,
          1
        ]
      },
      "points": {
        "mean": 2.9565217391304346,
        "ci": [
          1.9565217391304348,
          3.727272727272727
        ],
        "reports": 23,
        "fields": {
          "summary": {
            "mean": 1,
            "n": 21
          },
          "decided_by": {
            "mean": 1,
            "n": 4
          },
          "why": {
            "mean": 0.3333333333333333,
            "n": 12
          },
          "goal": {
            "mean": 0.6923076923076923,
            "n": 13
          },
          "metric": {
            "mean": 0.75,
            "n": 8
          },
          "direction": {
            "mean": 0.8181818181818182,
            "n": 11
          },
          "judge_by": {
            "mean": 1,
            "n": 3
          }
        }
      },
      "context": {
        "perRequest": 2058,
        "perSession": {
          "p25": 14525,
          "median": 16603,
          "p75": 21163
        },
        "sessions": 16
      },
      "costPerSession": 0.003232275,
      "costPerCapture": 0.00646455
    },
    {
      "model": "anthropic/claude-haiku-5.5",
      "run": "2026-10-31-00-51",
      "tools": "none",
      "prompt": "none",
      "skill": "full",
      "split": "all",
      "sessions": 16,
      "failed": 0,
      "right": {
        "k": 16,
        "n": 16,
        "rate": 1,
        "ci": [
          0.8063865817272802,
          1
        ]
      },
      "strict": {
        "k": 15,
        "n": 16,
        "rate": 0.9375,
        "ci": [
          0.7167073163197013,
          0.9888809426916687
        ]
      },
      "capture": {
        "k": 8,
        "n": 8,
        "rate": 1,
        "ci": [
          0.6755843804891231,
          1
        ]
      },
      "falseReports": {
        "k": 0,
        "n": 4,
        "rate": 0,
        "ci": [
          0,
          0.48990002040399916
        ]
      },
      "offRecord": {
        "k": 0,
        "n": 2,
        "rate": 0,
        "ci": [
          0,
          0.6576280471103807
        ]
      },
      "invented": {
        "k": 0,
        "n": 11,
        "rate": 0,
        "ci": [
          2.7755575615628914e-17,
          0.2588400172488141
        ]
      },
      "duplicates": {
        "k": 2,
        "n": 8,
        "rate": 0.25,
        "ci": [
          0.07147768885802763,
          0.5907301208974108
        ]
      },
      "points": {
        "mean": 3.4545454545454546,
        "ci": [
          2.272727272727273,
          4.75
        ],
        "reports": 11,
        "fields": {
          "summary": {
            "mean": 1,
            "n": 10
          },
          "goal": {
            "mean": 1,
            "n": 6
          },
          "why": {
            "mean": 1,
            "n": 6
          },
          "decided_by": {
            "mean": 1,
            "n": 2
          },
          "metric": {
            "mean": 0.3333333333333333,
            "n": 3
          },
          "direction": {
            "mean": 0.5,
            "n": 4
          },
          "judge_by": {
            "mean": 1,
            "n": 1
          }
        }
      },
      "context": {
        "perRequest": 5080,
        "perSession": {
          "p25": 25416,
          "median": 30584,
          "p75": 40929
        },
        "sessions": 16
      },
      "costPerSession": 0.0037868624999999995,
      "costPerCapture": 0.007573724999999999
    }
  ],
  "byKind": [
    {
      "run": "2026-10-31-00-51",
      "type": "not a decision",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "partial plan",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "made",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "deferred",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 0,
        "n": 5,
        "rate": 0,
        "ci": [
          0,
          0.43449149475208104
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "not a decision",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "someone else decided",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "someone else decided",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "off the record",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "deferred",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 0,
        "n": 5,
        "rate": 0,
        "ci": [
          0,
          0.43449149475208104
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "reversed",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "partial plan",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 1,
        "n": 5,
        "rate": 0.2,
        "ci": [
          0.03622316096978745,
          0.6244717358814613
        ]
      },
      "strict": {
        "k": 1,
        "n": 5,
        "rate": 0.2,
        "ci": [
          0.03622316096978745,
          0.6244717358814613
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "made",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "delegated",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "someone else decided",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "delegated",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 2,
        "n": 2,
        "rate": 1,
        "ci": [
          0.34237195288961925,
          1
        ]
      },
      "strict": {
        "k": 2,
        "n": 2,
        "rate": 1,
        "ci": [
          0.34237195288961925,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "someone else decided",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "delegated",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 2,
        "n": 3,
        "rate": 0.6666666666666666,
        "ci": [
          0.2076549551264879,
          0.9385096847238394
        ]
      },
      "strict": {
        "k": 2,
        "n": 3,
        "rate": 0.6666666666666666,
        "ci": [
          0.2076549551264879,
          0.9385096847238394
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "reversed",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "reversed",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "deferred",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "partial plan",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 0,
        "n": 4,
        "rate": 0,
        "ci": [
          0,
          0.48990002040399916
        ]
      },
      "strict": {
        "k": 0,
        "n": 4,
        "rate": 0,
        "ci": [
          0,
          0.48990002040399916
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "not a decision",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 2,
        "n": 2,
        "rate": 1,
        "ci": [
          0.34237195288961925,
          1
        ]
      },
      "strict": {
        "k": 2,
        "n": 2,
        "rate": 1,
        "ci": [
          0.34237195288961925,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "made",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 3,
        "n": 3,
        "rate": 1,
        "ci": [
          0.43849391955098227,
          1
        ]
      },
      "strict": {
        "k": 3,
        "n": 3,
        "rate": 1,
        "ci": [
          0.43849391955098227,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "off the record",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "deferred",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 3,
        "n": 5,
        "rate": 0.6,
        "ci": [
          0.23071993220883708,
          0.8823817688407467
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "not a decision",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "reversed",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "partial plan",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 4,
        "n": 5,
        "rate": 0.8,
        "ci": [
          0.3755282641185388,
          0.9637768390302125
        ]
      },
      "strict": {
        "k": 4,
        "n": 5,
        "rate": 0.8,
        "ci": [
          0.3755282641185388,
          0.9637768390302125
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "off the record",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "off the record",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 0,
        "n": 2,
        "rate": 0,
        "ci": [
          0,
          0.6576280471103807
        ]
      },
      "strict": {
        "k": 0,
        "n": 2,
        "rate": 0,
        "ci": [
          0,
          0.6576280471103807
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "delegated",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 2,
        "n": 4,
        "rate": 0.5,
        "ci": [
          0.15003570882017148,
          0.8499642911798285
        ]
      },
      "strict": {
        "k": 2,
        "n": 4,
        "rate": 0.5,
        "ci": [
          0.15003570882017148,
          0.8499642911798285
        ]
      }
    },
    {
      "run": "2026-10-31-00-51",
      "type": "made",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "made",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "someone else decided",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "reversed",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "reversed",
      "phrasing": "blunt",
      "skill": "compact",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "someone else decided",
      "phrasing": "multi-turn",
      "skill": "compact",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "partial plan",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 0,
        "n": 5,
        "rate": 0,
        "ci": [
          0,
          0.43449149475208104
        ]
      },
      "strict": {
        "k": 0,
        "n": 5,
        "rate": 0,
        "ci": [
          0,
          0.43449149475208104
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "made",
      "phrasing": "blunt",
      "skill": "compact",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "partial plan",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "partial plan",
      "phrasing": "multi-turn",
      "skill": "compact",
      "right": {
        "k": 4,
        "n": 5,
        "rate": 0.8,
        "ci": [
          0.3755282641185388,
          0.9637768390302125
        ]
      },
      "strict": {
        "k": 4,
        "n": 5,
        "rate": 0.8,
        "ci": [
          0.3755282641185388,
          0.9637768390302125
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "reversed",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "reversed",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "reversed",
      "phrasing": "multi-turn",
      "skill": "compact",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "partial plan",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 4,
        "n": 5,
        "rate": 0.8,
        "ci": [
          0.3755282641185388,
          0.9637768390302125
        ]
      },
      "strict": {
        "k": 4,
        "n": 5,
        "rate": 0.8,
        "ci": [
          0.3755282641185388,
          0.9637768390302125
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "someone else decided",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "partial plan",
      "phrasing": "blunt",
      "skill": "compact",
      "right": {
        "k": 0,
        "n": 5,
        "rate": 0,
        "ci": [
          0,
          0.43449149475208104
        ]
      },
      "strict": {
        "k": 0,
        "n": 5,
        "rate": 0,
        "ci": [
          0,
          0.43449149475208104
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "made",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "reversed",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "made",
      "phrasing": "multi-turn",
      "skill": "compact",
      "right": {
        "k": 3,
        "n": 3,
        "rate": 1,
        "ci": [
          0.43849391955098227,
          1
        ]
      },
      "strict": {
        "k": 3,
        "n": 3,
        "rate": 1,
        "ci": [
          0.43849391955098227,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "partial plan",
      "phrasing": "blunt",
      "skill": "none",
      "right": {
        "k": 0,
        "n": 4,
        "rate": 0,
        "ci": [
          0,
          0.48990002040399916
        ]
      },
      "strict": {
        "k": 0,
        "n": 4,
        "rate": 0,
        "ci": [
          0,
          0.48990002040399916
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "made",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      },
      "strict": {
        "k": 4,
        "n": 4,
        "rate": 1,
        "ci": [
          0.5100999795960008,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "someone else decided",
      "phrasing": "blunt",
      "skill": "compact",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "someone else decided",
      "phrasing": "blunt",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "someone else decided",
      "phrasing": "multi-turn",
      "skill": "full",
      "right": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      },
      "strict": {
        "k": 5,
        "n": 5,
        "rate": 1,
        "ci": [
          0.5655085052479191,
          1
        ]
      }
    },
    {
      "run": "2026-10-31-02-16",
      "type": "made",
      "phrasing": "multi-turn",
      "skill": "none",
      "right": {
        "k": 3,
        "n": 3,
        "rate": 1,
        "ci": [
          0.43849391955098227,
          1
        ]
      },
      "strict": {
        "k": 3,
        "n": 3,
        "rate": 1,
        "ci": [
          0.43849391955098227,
          1
        ]
      }
    }
  ],
  "changes": [
    {
      "date": "2026-10-31",
      "text": "First run. Five models ran every conversation twice: once with ZeroWidth's tools only, and once with ZeroWidth's tools and instructions.",
      "models": [
        "anthropic/claude-sonnet-5.5",
        "anthropic/claude-haiku-5.5",
        "google/gemini-3.8-flash",
        "deepseek/deepseek-v4.1-flash",
        "openai/gpt-6-luna"
      ]
    },
    {
      "date": "2026-10-31",
      "text": "Second run, on the eight conversations that hold a decision. ZeroWidth now tells the assistant when a decision is already recorded, so the assistant stops recording the same decision twice. A third setup joined: tools and short instructions, about a fifth the length of the full instructions.",
      "models": [
        "anthropic/claude-sonnet-5.5",
        "anthropic/claude-haiku-5.5",
        "google/gemini-3.8-flash",
        "deepseek/deepseek-v4.1-flash",
        "openai/gpt-6-luna"
      ]
    }
  ]
}
