Calibration

V* → We-Math

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

We-Math is estimated from V* with a linear curve fitted on 5 models measured on both: y = 0.5159·x + 0.3586, R² = 0.41, cross-validated error 8.5 pp. It is used for 0 estimates.

Fitted on

Model	V*	We-Math
Claude Opus 4.5	67.0%	70.0%
Gemini 3 Pro	88.0%	86.9%
GPT-5.2	75.9%	79.0%
LFM3-VL-2B	86.4%	68.6%
Qwen3.5 397B	95.8%	87.9%