AdvancedMathBench - VerifierBench Rough: leaderboard

Metric: Rough balanced F1 (%): binary proof-validity verdicts only, 888 model-generated proof trajectories with expert labels; balanced F1 is the harmonic mean of the true positive and true negative rates; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 11 models tracked.

Top models

#ModelScore
1GPT-5.5 (xHigh)76
2GPT-5.5 (High)76
3DeepSeek V4 Pro (Max)74.1
4GLM-5.2 (Max)73.9
5Qwen 3.5 397B A17B68.9
6GPT-5.2 (xHigh)65.9
7Gemini 3.1 Pro (Preview) (High)64.4
8Claude Opus 4.8 (Max)54.4

Interactive version: theaggregate.ai/benchmark?slug=advancedmathbench-verifierbench-rough · How It Works · Data refreshed daily, snapshot 2026-09-29.