Calibration

We-Math → SimpleVQA

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

SimpleVQA is estimated from We-Math with a Michaelis–Menten curve fitted on 6 models measured on both: y = 2.0000·x / (1.77544 + x), R² = 0.37, cross-validated error 12.4 pp. It is used for 0 estimates.

Fitted on

Model	We-Math	SimpleVQA
Claude Opus 4.5	70.0%	65.7%
Gemini 3 Pro	86.9%	73.2%
GPT-5.2	79.0%	55.8%
Kimi K2.5	84.7%	71.2%
LFM3-VL-2B	68.6%	36.3%
Qwen3.5 397B	87.9%	67.1%