ArxivMath — leaderboard
MathArena ArxivMath final-answer research-math benchmark slice from the March and April 2026 releases, as reported in Anthropic's Claude Opus 4.8 system card.
Metric: Score (self-reported). Source: benchmarklist.com. Status: saturation imminent. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 | 78.6 |
| 2 | Claude Mythos 5 | 78.52 |
| 3 | GPT-5.5 | 72.2 |
| 4 | Claude Sonnet 5 | 72.2 |
| 5 | Claude Opus 4.8 | 71 |
| 6 | Claude Mythos Preview | 68.7 |
| 7 | Gemini 3.1 Pro (Preview) | 64.79 |
| 8 | Claude Opus 4.7 | 62.7 |
| 9 | Claude Opus 4.6 | 60.6 |
| 10 | Gemini 3.5 Flash | 47.7 |
Interactive version: theaggregate.ai/benchmark?slug=arxivmath · How the rankings work · Data refreshed daily, snapshot 2026-07-22.