Calibration

MMMU → RealWorldQA

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

RealWorldQA is estimated from MMMU with a offset logistic curve fitted on 5 models measured on both: y = 0.5842 + (0.8470 − 0.5842) / (1 + exp(−55.52·(x − 0.4798))), R² = 1.00, cross-validated error 0.8 pp. It is used for 2 estimates.

Estimated model	MMMU	RealWorldQA	Source
Command A+	75.1%	84.7%	estimated ± 0.8 pp, medium confidence
Nemotron 3 Nano Omni 30B A3B	70.8%	84.7%	estimated ± 0.8 pp, medium confidence