Calibration
CritPt → GPQA Diamond
GPQA Diamond is estimated from CritPt with a Michaelis–Menten curve fitted on 16 models measured on both: y = 0.9508·x / (0.00309 + x), R² = 0.71, cross-validated error 2.3 pp. It is used for 1 estimate.
| Estimated model | CritPt | GPQA Diamond | Source |
|---|---|---|---|
| GPT-6.1 Sol (xhigh) | 31.7% | 94.2% | estimated ± 2.3 pp, high confidence |