Calibration
SuperGPQA → MMLU-Redux
MMLU-Redux is estimated from SuperGPQA with a Michaelis–Menten curve fitted on 8 models measured on both: y = 1.1442·x / (0.14547 + x), R² = 0.99, cross-validated error 0.9 pp. It is used for 3 estimates.
| Estimated model | SuperGPQA | MMLU-Redux | Source |
|---|---|---|---|
| Qwen 3.6 Max (preview) | 73.9% | 95.6% | estimated ± 0.9 pp, medium confidence |
| Seed 2.1 Pro | 70.8% | 94.9% | estimated ± 0.9 pp, high confidence |
| Seed 2.1 Turbo | 67.4% | 94.1% | estimated ± 0.9 pp, high confidence |