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

#ModelScore
1Claude Fable 578.6
2Claude Mythos 578.52
3GPT-5.572.2
4Claude Sonnet 572.2
5Claude Opus 4.871
6Claude Mythos Preview68.7
7Gemini 3.1 Pro (Preview)64.79
8Claude Opus 4.762.7
9Claude Opus 4.660.6
10Gemini 3.5 Flash47.7

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