Calibration
MMMU-Pro → MMMU
MMMU is estimated from MMMU-Pro with a Michaelis–Menten + offset curve fitted on 5 models measured on both: y = 0.1218 + 2.0000·x / (1.37916 + x), R² = 1.00, cross-validated error 1.1 pp. It is used for 8 estimates.
| Estimated model | MMMU-Pro | MMMU | Source |
|---|---|---|---|
| dots3-note Preview | 79.1% | 85.1% | estimated ± 1.1 pp, low confidence |
| GPT-5.2 | 79.5% | 85.3% | estimated ± 1.1 pp, low confidence |
| Grok 4.20 | 75.2% | 82.7% | estimated ± 1.1 pp, medium confidence |
| MiMo-V2.5 | 77.9% | 84.4% | estimated ± 1.1 pp, medium confidence |
| Pareto 26.9 | 78.0% | 84.4% | estimated ± 1.1 pp, medium confidence |
| Qwen3.8 Max | 82.3% | 86.9% | estimated ± 1.1 pp, low confidence |
| Seed 2.1 Pro | 81.6% | 86.5% | estimated ± 1.1 pp, low confidence |
| Seed 2.1 Turbo | 80.1% | 85.7% | estimated ± 1.1 pp, low confidence |