Calibration

SimpleVQA → MMMU-Pro

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

MMMU-Pro is estimated from SimpleVQA with a Hill curve fitted on 10 models measured on both: y = 0.0000 + (0.8291 − 0.0000)·x^6.00 / (0.38609^6.00 + x^6.00), R² = 0.98, cross-validated error 2.4 pp. It is used for 1 estimate.

Estimated model	SimpleVQA	MMMU-Pro	Source
Step 3.7 Flash	79.2%	81.8%	estimated ± 2.4 pp, high confidence