MemoryArena - Math Formal Reasoning Success Rate: leaderboard
Metric: Task success rate (%) on sequential math formal reasoning (40 groups of 2-16 lemma subtasks; success means the main problem is solved); the model is the task agent and keeps its full interaction history in context (no external memory); printed as fractions and shown times 100; higher is better. Source: arxiv.org. Saturation forecast: Around September 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.6 | 37 | #85 |
| 2 | Claude Sonnet 4.5 | 20 | #138 |
| 3 | Gemini 3 Flash | 17 | #93 |
| 4 | GPT-4.1 Mini | 13 | #346 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=memoryarena-math-formal-reasoning-success-rate · How It Works · Data refreshed daily, snapshot 2026-10-11.