Snapshot: 2026-09-30T20:54:24.517Z
Rank method: modelcap-index-v4.12-fused-launch-placement
Evidence admission: modelcap-score-v10-best-supported-configuration
Ranked models: 418
Measured scores from benchmark rows: 230 of 230 rebuilt within 0.15 points
Family scores from benchmark rows: 562 of 562 rebuilt within 0.15 points
Combined scores from their estimates: 188 of 188 rebuilt within 0.15 points
Measured estimates inside combined scores: 11 of 11 rebuilt within 0.15 points
Scores without rebuildable inputs: 0
Current board: 279 of 279 positions reproduced
All versions: 418 of 418 positions reproduced
Rank  Score  Rebuilt  Basis      Confidence  Model
   1   91.9    91.90  measured         20.0  Claude Sonnet 5.5
   2   88.4    88.48  measured         65.4  Claude Opus 5.5
   3   88.2    88.19  measured         23.0  GPT-6.1 Sol
   4   86.9    86.89  measured         83.2  Claude Fable 5.1
   5   84.7    84.66  measured         82.4  Muse Spark 1.3
   6   83.6    83.57  measured         81.1  GPT-6 Astra
   7   83.3    83.30  measured         20.0  Qwen3.8 Max (0902)
   8   83.2    83.24  measured         70.8  MiMo-V2.6-Pro
   9   82.7    82.67  measured         88.6  Gemini 3.8 Flash
  10   82.0    81.95  measured         89.0  Kimi K3
Worked example: Claude Sonnet 5.5
  general    Artificial Analysis Intelligence Index: 91.9 x share 0.25 x confidence 1.000
  general family: 91.90
  rebuilt score: 91.90
  published score: 91.9
@misc{modelcap_index,
  author = {{ModelCap}},
  title = {Language model rankings dataset},
  year = {2026},
  url = {https://modelcap.ai/data},
  note = {Snapshot 2026-09-30T20:54:24.517Z, rank method modelcap-index-v4.12-fused-launch-placement}
}