Calibration
MMMU-Pro → AA-MMMU-Pro
AA-MMMU-Pro is estimated from MMMU-Pro with a Hill curve fitted on 32 models measured on both: y = 0.5775 + (1.2000 − 0.5775)·x^6.00 / (0.90276^6.00 + x^6.00), R² = 0.46, cross-validated error 4.9 pp. It is used for 4 estimates.
| Estimated model | MMMU-Pro | AA-MMMU-Pro | Source |
|---|---|---|---|
| GPT-5.2 | 79.5% | 77.6% | estimated ± 4.9 pp, medium confidence |
| Interfaze Beta | 71.1% | 69.7% | estimated ± 4.9 pp, medium confidence |
| MiMo-V2.5 | 77.9% | 75.9% | estimated ± 4.9 pp, medium confidence |
| Pareto 26.9 | 78.0% | 76.0% | estimated ± 4.9 pp, medium confidence |