Calibration
Vals GPQA Diamond → AA-GPQA Diamond
AA-GPQA Diamond is estimated from Vals GPQA Diamond with a offset logistic curve fitted on 47 models measured on both: y = 0.6871 + (0.9480 − 0.6871) / (1 + exp(−20.06·(x − 0.8103))), R² = 0.78, cross-validated error 4.2 pp. It is used for 7 estimates.
| Estimated model | Vals GPQA Diamond | AA-GPQA Diamond | Source |
|---|---|---|---|
| Claude Haiku 4.5 | 72.2% | 72.5% | estimated ± 4.2 pp, high confidence |
| Gemini 3.1 Flash-Lite | 81.1% | 81.9% | estimated ± 4.2 pp, high confidence |
| GLM-4.5 | 72.2% | 72.5% | estimated ± 4.2 pp, high confidence |
| Laguna M.1 | 27.0% | 68.7% | estimated ± 4.2 pp, medium confidence |
| Laguna XS.2 | 55.1% | 68.9% | estimated ± 4.2 pp, high confidence |
| MiMo-V2.5 | 81.6% | 82.5% | estimated ± 4.2 pp, high confidence |
| Qwen3.5 Flash | 82.8% | 84.1% | estimated ± 4.2 pp, high confidence |