{
 "banc": "ChaserBench v1",
 "seed": 42,
 "alpha": 0.05,
 "date": "2026-07-17 02:01:16",
 "corpus": {
  "fichier": "bench\\corpus_2026Q3.jsonl",
  "sha256": "c35ce8ff1db79a88dbe9ac4c3ba2894d802ad955b49883c1d89cc04bed3b6d9c",
  "n_taches": 30
 },
 "perimetre_honnete": "v1 : coût/qualité/latence sur tâches vérifiables en un tour ; les tool calls sont HORS périmètre (mesurés par l'outil MCP attester, v2 prévue)",
 "bras": {
  "reference": {
   "cout_total_usd": 0.10133,
   "reussite": {
    "n": 30,
    "centre": 1.0,
    "borne_basse": 0.7489,
    "borne_haute": 1.0
   },
   "latence_mediane_s": 13.07
  },
  "chaser": {
   "cout_total_usd": 0.04328,
   "reussite": {
    "n": 30,
    "centre": 0.9667,
    "borne_basse": 0.7025,
    "borne_haute": 1.0
   },
   "latence_mediane_s": 11.75,
   "escalades": 0
  }
 },
 "paires": {
  "gain_cout_relatif": {
   "n": 30,
   "moyenne": 0.6354,
   "borne_basse": 0.134,
   "borne_haute": 1.0
  },
  "delta_qualite_ref_moins_chaser": {
   "n": 30,
   "moyenne": 0.0334,
   "borne_basse": -0.4654,
   "borne_haute": 0.532
  },
  "lecture": "gain_cout_relatif > 0 = chaser moins cher à tâche égale ; delta_qualite proche de 0 = qualité préservée (IC anytime-valid, peeking permis)"
 },
 "par_tache": [
  {
   "id": "m10",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.007052,
    "tokens_in": 46,
    "tokens_out": 696,
    "latence_s": 13.07,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.003784,
    "tokens_in": 34,
    "tokens_out": 750,
    "latence_s": 5.31,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "m05",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.001694,
    "tokens_in": 47,
    "tokens_out": 160,
    "latence_s": 3.26,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000312,
    "tokens_in": 37,
    "tokens_out": 55,
    "latence_s": 1.95,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "m01",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.005398,
    "tokens_in": 39,
    "tokens_out": 532,
    "latence_s": 6.52,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.003283,
    "tokens_in": 33,
    "tokens_out": 650,
    "latence_s": 6.51,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c07",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.00811,
    "tokens_in": 55,
    "tokens_out": 800,
    "latence_s": 9.39,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.004042,
    "tokens_in": 42,
    "tokens_out": 800,
    "latence_s": 7.3,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c03",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.008122,
    "tokens_in": 61,
    "tokens_out": 800,
    "latence_s": 9.25,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.004048,
    "tokens_in": 48,
    "tokens_out": 800,
    "latence_s": 7.38,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t07",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000318,
    "tokens_in": 44,
    "tokens_out": 23,
    "latence_s": 4.65,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000146,
    "tokens_in": 36,
    "tokens_out": 22,
    "latence_s": 4.07,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t06",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000334,
    "tokens_in": 42,
    "tokens_out": 25,
    "latence_s": 5.05,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000137,
    "tokens_in": 32,
    "tokens_out": 21,
    "latence_s": 4.64,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "m03",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.004776,
    "tokens_in": 33,
    "tokens_out": 471,
    "latence_s": 9.68,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.002791,
    "tokens_in": 26,
    "tokens_out": 553,
    "latence_s": 8.52,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "m02",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.007046,
    "tokens_in": 33,
    "tokens_out": 698,
    "latence_s": 12.15,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.002411,
    "tokens_in": 26,
    "tokens_out": 477,
    "latence_s": 8.71,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "m06",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.003748,
    "tokens_in": 39,
    "tokens_out": 367,
    "latence_s": 9.25,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.001066,
    "tokens_in": 31,
    "tokens_out": 207,
    "latence_s": 7.63,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t10",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000948,
    "tokens_in": 44,
    "tokens_out": 86,
    "latence_s": 7.94,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": false,
    "cout_usd": 0.000141,
    "tokens_in": 31,
    "tokens_out": 22,
    "latence_s": 7.06,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c06",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.003044,
    "tokens_in": 27,
    "tokens_out": 299,
    "latence_s": 9.53,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.001403,
    "tokens_in": 23,
    "tokens_out": 276,
    "latence_s": 8.52,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c10",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.003442,
    "tokens_in": 36,
    "tokens_out": 337,
    "latence_s": 11.35,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000488,
    "tokens_in": 28,
    "tokens_out": 92,
    "latence_s": 8.99,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c02",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.004482,
    "tokens_in": 36,
    "tokens_out": 441,
    "latence_s": 11.83,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.001461,
    "tokens_in": 36,
    "tokens_out": 285,
    "latence_s": 10.35,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "m07",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.005382,
    "tokens_in": 41,
    "tokens_out": 530,
    "latence_s": 14.23,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.003357,
    "tokens_in": 32,
    "tokens_out": 665,
    "latence_s": 13.15,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c08",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000708,
    "tokens_in": 74,
    "tokens_out": 56,
    "latence_s": 10.85,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000185,
    "tokens_in": 60,
    "tokens_out": 25,
    "latence_s": 11.0,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t02",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000656,
    "tokens_in": 43,
    "tokens_out": 57,
    "latence_s": 11.55,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000141,
    "tokens_in": 36,
    "tokens_out": 21,
    "latence_s": 11.75,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "m04",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.008074,
    "tokens_in": 37,
    "tokens_out": 800,
    "latence_s": 18.41,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.004027,
    "tokens_in": 27,
    "tokens_out": 800,
    "latence_s": 17.19,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "m09",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000898,
    "tokens_in": 49,
    "tokens_out": 80,
    "latence_s": 13.18,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000165,
    "tokens_in": 40,
    "tokens_out": 25,
    "latence_s": 12.49,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t03",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.001502,
    "tokens_in": 31,
    "tokens_out": 144,
    "latence_s": 15.04,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000266,
    "tokens_in": 26,
    "tokens_out": 48,
    "latence_s": 13.33,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "m08",
   "categorie": "moyenne",
   "reference": {
    "reussite": true,
    "cout_usd": 0.004204,
    "tokens_in": 42,
    "tokens_out": 412,
    "latence_s": 17.02,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000848,
    "tokens_in": 33,
    "tokens_out": 163,
    "latence_s": 14.66,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c09",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.008106,
    "tokens_in": 53,
    "tokens_out": 800,
    "latence_s": 21.2,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.003971,
    "tokens_in": 41,
    "tokens_out": 786,
    "latence_s": 18.66,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t05",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000234,
    "tokens_in": 37,
    "tokens_out": 16,
    "latence_s": 14.9,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 8.7e-05,
    "tokens_in": 27,
    "tokens_out": 12,
    "latence_s": 14.54,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c05",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.008064,
    "tokens_in": 32,
    "tokens_out": 800,
    "latence_s": 22.15,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.003187,
    "tokens_in": 27,
    "tokens_out": 632,
    "latence_s": 18.51,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t08",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.001216,
    "tokens_in": 33,
    "tokens_out": 115,
    "latence_s": 16.64,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000275,
    "tokens_in": 30,
    "tokens_out": 49,
    "latence_s": 15.96,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t09",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000268,
    "tokens_in": 39,
    "tokens_out": 19,
    "latence_s": 16.99,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000116,
    "tokens_in": 31,
    "tokens_out": 17,
    "latence_s": 16.62,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c04",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000658,
    "tokens_in": 54,
    "tokens_out": 55,
    "latence_s": 17.66,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000323,
    "tokens_in": 43,
    "tokens_out": 56,
    "latence_s": 17.44,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t01",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.00066,
    "tokens_in": 35,
    "tokens_out": 59,
    "latence_s": 18.53,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 9.9e-05,
    "tokens_in": 29,
    "tokens_out": 14,
    "latence_s": 17.66,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "t04",
   "categorie": "triviale",
   "reference": {
    "reussite": true,
    "cout_usd": 0.00125,
    "tokens_in": 45,
    "tokens_out": 116,
    "latence_s": 19.09,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000123,
    "tokens_in": 33,
    "tokens_out": 18,
    "latence_s": 18.66,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  },
  {
   "id": "c01",
   "categorie": "complexe",
   "reference": {
    "reussite": true,
    "cout_usd": 0.000934,
    "tokens_in": 57,
    "tokens_out": 82,
    "latence_s": 19.18,
    "modele_final": "claude-sonnet-5",
    "escalades": 0
   },
   "chaser": {
    "reussite": true,
    "cout_usd": 0.000598,
    "tokens_in": 48,
    "tokens_out": 110,
    "latence_s": 19.84,
    "modele_final": "claude-haiku-4-5",
    "escalades": 0
   }
  }
 ],
 "scelle_merkle": true
}