ProofBench — leaderboard

ProofBench evaluates model capability on math tasks from the linked upstream source with Score as the primary reported metric.

Metric: Score (self-reported). Source: vals.ai. Status: saturation imminent. 37 models tracked.

Top models

#ModelScore
1Claude Fable 5 maxAnthropic77
2Aristotle Harmonic71
3Claude Opus 4.8 maxAnthropic69
4GPT-5.4 x-highOpenAI56
5Claude Opus 4.7 maxAnthropic54
6Claude Opus 4.6 maxAnthropic50
7GPT-5.5 x-highOpenAI50
8Claude Sonnet 4.6 maxAnthropic45
9Claude Opus 4.5 highAnthropic36
10Gemini 3.5 Flash highGoogle29

Interactive version: theaggregate.ai/benchmark?slug=proofbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.