Calibration
GPQA Diamond → MMLU-Redux
MMLU-Redux is estimated from GPQA Diamond with a linear curve fitted on 7 models measured on both: y = 0.3348·x + 0.6322, R² = 0.93, cross-validated error 3.0 pp. It is used for 4 estimates.
| Estimated model | GPQA Diamond | MMLU-Redux | Source |
|---|---|---|---|
| Grok 4.20 | 88.5% | 92.8% | estimated ± 3.0 pp, medium confidence |
| LLaDA2.2-mini | 44.4% | 78.1% | estimated ± 3.0 pp, medium confidence |
| Pareto 26.10 Preview | 92.4% | 94.1% | estimated ± 3.0 pp, medium confidence |
| Beam | 90.5% | 93.5% | estimated ± 3.0 pp, medium confidence |