{
  "generated_from": "benchmarks/results/*.jsonl",
  "note": "chainwalk omitted: n=2 for every model, never a usable sample",
  "suites": [
    "nullschema",
    "worldview",
    "grounding",
    "epistemics",
    "toolcall",
    "contract",
    "synthesis",
    "enrichment",
    "screening",
    "tagger",
    "tagger_nullable"
  ],
  "models": [
    {
      "id": "bonsai-8b-1bit",
      "name": "Bonsai 8B 1-bit",
      "tps": 68.5,
      "gb": 1.1,
      "scores": {
        "grounding": {
          "pct": 55.2,
          "n": 125
        },
        "screening": {
          "pct": 86.7,
          "n": 60
        },
        "nullschema": {
          "pct": 19.4,
          "n": 72
        },
        "worldview": {
          "pct": 40.0,
          "n": 40
        },
        "contract": {
          "pct": 20.8,
          "n": 48
        },
        "tagger_nullable": {
          "pct": 86.3,
          "n": 240
        },
        "enrichment": {
          "pct": 83.0,
          "n": 54
        },
        "epistemics": {
          "pct": 80.8,
          "n": 48
        },
        "synthesis": {
          "pct": 65.0,
          "n": 60
        },
        "toolscale": {
          "pct": 100.0,
          "n": 40
        },
        "tools": {
          "pct": 96.0,
          "n": 15
        },
        "tagger": {
          "pct": 80.3,
          "n": 240
        },
        "toolcall": {
          "pct": 0.0,
          "n": 60
        }
      }
    },
    {
      "id": "ornith-35b-a3b-q40",
      "name": "Ornith 35B-A3B",
      "tps": 53.4,
      "gb": 21.7,
      "scores": {
        "grounding": {
          "pct": 76.0,
          "n": 125
        },
        "toolcall": {
          "pct": 80.0,
          "n": 60
        },
        "contract": {
          "pct": 66.7,
          "n": 48
        },
        "worldview": {
          "pct": 85.0,
          "n": 40
        },
        "toolscale": {
          "pct": 100.0,
          "n": 40
        },
        "nullschema": {
          "pct": 36.1,
          "n": 72
        }
      }
    },
    {
      "id": "btl-4-q4km",
      "name": "BTL-4 Q4_K_M",
      "tps": 48.5,
      "gb": 20.5,
      "scores": {
        "worldview": {
          "pct": 55.0,
          "n": 40
        },
        "tools": {
          "pct": 100.0,
          "n": 15
        },
        "nullschema": {
          "pct": 44.4,
          "n": 72
        },
        "toolscale": {
          "pct": 100.0,
          "n": 40
        },
        "tagger": {
          "pct": 96.4,
          "n": 240
        },
        "synthesis": {
          "pct": 83.3,
          "n": 60
        },
        "screening": {
          "pct": 91.7,
          "n": 60
        },
        "tagger_nullable": {
          "pct": 94.1,
          "n": 164
        },
        "contract": {
          "pct": 70.8,
          "n": 48
        },
        "epistemics": {
          "pct": 100.0,
          "n": 48
        },
        "toolcall": {
          "pct": 93.3,
          "n": 60
        },
        "enrichment": {
          "pct": 98.8,
          "n": 54
        },
        "grounding": {
          "pct": 76.8,
          "n": 125
        }
      }
    },
    {
      "id": "qwen3.8-9b-distill",
      "name": "Qwen3.8-9B distill",
      "tps": 30.8,
      "gb": 9.8,
      "scores": {
        "enrichment": {
          "pct": 95.9,
          "n": 54
        },
        "contract": {
          "pct": 25.0,
          "n": 48
        },
        "worldview": {
          "pct": 40.0,
          "n": 40
        },
        "synthesis": {
          "pct": 83.3,
          "n": 60
        },
        "epistemics": {
          "pct": 95.8,
          "n": 48
        },
        "tagger_nullable": {
          "pct": 96.3,
          "n": 240
        },
        "toolscale": {
          "pct": 100.0,
          "n": 40
        },
        "nullschema": {
          "pct": 18.1,
          "n": 72
        },
        "screening": {
          "pct": 91.7,
          "n": 60
        },
        "tools": {
          "pct": 100.0,
          "n": 15
        },
        "grounding": {
          "pct": 67.2,
          "n": 125
        },
        "toolcall": {
          "pct": 0.0,
          "n": 60
        },
        "tagger": {
          "pct": 94.4,
          "n": 240
        }
      }
    },
    {
      "id": "btl-4",
      "name": "BTL-4 IQ2_XXS",
      "tps": 23.6,
      "gb": 10.0,
      "scores": {
        "toolscale": {
          "pct": 100.0,
          "n": 40
        },
        "contract": {
          "pct": 64.6,
          "n": 48
        },
        "grounding": {
          "pct": 76.8,
          "n": 125
        },
        "screening": {
          "pct": 95.0,
          "n": 60
        },
        "synthesis": {
          "pct": 80.0,
          "n": 60
        },
        "epistemics": {
          "pct": 100.0,
          "n": 48
        },
        "tools": {
          "pct": 100.0,
          "n": 15
        },
        "nullschema": {
          "pct": 36.1,
          "n": 72
        },
        "tagger_nullable": {
          "pct": 96.2,
          "n": 240
        },
        "enrichment": {
          "pct": 98.7,
          "n": 54
        },
        "toolcall": {
          "pct": 76.7,
          "n": 60
        },
        "tagger": {
          "pct": 96.5,
          "n": 240
        },
        "worldview": {
          "pct": 62.5,
          "n": 40
        }
      }
    },
    {
      "id": "bonsai-27b-1bit",
      "name": "Bonsai 27B 1-bit",
      "tps": 20.0,
      "gb": 3.5,
      "scores": {
        "contract": {
          "pct": 25.0,
          "n": 48
        },
        "screening": {
          "pct": 96.7,
          "n": 60
        },
        "tools": {
          "pct": 100.0,
          "n": 15
        },
        "worldview": {
          "pct": 40.0,
          "n": 40
        },
        "enrichment": {
          "pct": 95.1,
          "n": 54
        },
        "nullschema": {
          "pct": 18.1,
          "n": 72
        },
        "tagger": {
          "pct": 91.7,
          "n": 240
        },
        "tagger_nullable": {
          "pct": 95.0,
          "n": 240
        },
        "toolscale": {
          "pct": 100.0,
          "n": 40
        },
        "epistemics": {
          "pct": 87.5,
          "n": 48
        },
        "toolcall": {
          "pct": 100.0,
          "n": 60
        },
        "grounding": {
          "pct": 67.2,
          "n": 125
        },
        "synthesis": {
          "pct": 35.0,
          "n": 60
        }
      }
    },
    {
      "id": "ternary-bonsai-27b",
      "name": "Bonsai 27B ternary",
      "tps": 17.3,
      "gb": 7.2,
      "scores": {
        "grounding": {
          "pct": 68.8,
          "n": 125
        },
        "toolcall": {
          "pct": 100.0,
          "n": 60
        },
        "tagger": {
          "pct": 96.2,
          "n": 240
        },
        "screening": {
          "pct": 93.3,
          "n": 60
        },
        "epistemics": {
          "pct": 85.4,
          "n": 48
        },
        "tools": {
          "pct": 100.0,
          "n": 15
        },
        "tagger_nullable": {
          "pct": 97.9,
          "n": 240
        },
        "nullschema": {
          "pct": 27.8,
          "n": 72
        },
        "worldview": {
          "pct": 42.5,
          "n": 40
        },
        "toolscale": {
          "pct": 100.0,
          "n": 40
        },
        "synthesis": {
          "pct": 66.7,
          "n": 60
        },
        "contract": {
          "pct": 75.0,
          "n": 48
        },
        "enrichment": {
          "pct": 88.4,
          "n": 54
        }
      }
    }
  ]
}