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
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 maxAnthropic | 77 |
| 2 | Aristotle Harmonic | 71 |
| 3 | Claude Opus 4.8 maxAnthropic | 69 |
| 4 | GPT-5.4 x-highOpenAI | 56 |
| 5 | Claude Opus 4.7 maxAnthropic | 54 |
| 6 | Claude Opus 4.6 maxAnthropic | 50 |
| 7 | GPT-5.5 x-highOpenAI | 50 |
| 8 | Claude Sonnet 4.6 maxAnthropic | 45 |
| 9 | Claude Opus 4.5 highAnthropic | 36 |
| 10 | Gemini 3.5 Flash highGoogle | 29 |
Interactive version: theaggregate.ai/benchmark?slug=proofbench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.