benchgap
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 modelMMMU-ProAA-MMMU-ProSource
GPT-5.279.5%77.6%estimated ± 4.9 pp, medium confidence
Interfaze Beta71.1%69.7%estimated ± 4.9 pp, medium confidence
MiMo-V2.577.9%75.9%estimated ± 4.9 pp, medium confidence
Pareto 26.978.0%76.0%estimated ± 4.9 pp, medium confidence