Calibration

We-Math → MMMU-Pro

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap
License: https://creativecommons.org/licenses/by-nc/4.0/

MMMU-Pro is estimated from We-Math with a offset logistic curve fitted on 6 models measured on both: y = 0.0018 + (0.7950 − 0.0018) / (1 + exp(−126.48·(x − 0.6836))), R² = 1.00, cross-validated error 11.3 pp. It is used for 0 estimates.

Fitted on

Model	We-Math	MMMU-Pro
Claude Opus 4.5	70.0%	70.6%
Gemini 3 Pro	86.9%	81.0%
GPT-5.2	79.0%	79.5%
Kimi K2.5	84.7%	78.5%
LFM3-VL-2B	68.6%	45.7%
Qwen3.5 397B	87.9%	79.0%