Vellum - MATH: leaderboard
Vellum's independently-run MATH evaluation. Competition-level math problems across algebra, geometry, number theory, and more.
Metric: Accuracy (%). Source: www.vellum.ai. Saturation forecast: Estimated already saturated. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Kimi K2.5 | 98 | #144 |
| 2 | O3 Mini | 97.9 | #272 |
| 3 | Claude Sonnet 4.6 | 97.8 | #90 |
| 4 | Claude Opus 4.6 | 97.6 | #63 |
| 5 | DeepSeek R1 | 97.3 | #251 |
| 6 | O1 | 96.4 | #183 |
| 7 | Claude 3.7 Sonnet (Thinking) | 96.2 | #249 (Claude 3.7 Sonnet) |
| 8 | DeepSeek V3 (0324) | 94 | #340 |
| 9 | O1 Mini | 90 | #323 |
| 10 | Gemini 2.0 Flash | 89.7 | #342 |
| 11 | Gemma 3 27B | 89 | #607 |
| 12 | Claude 3.7 Sonnet | 82.2 | #249 |
| 13 | Claude 3.5 Sonnet | 78 | #350 |
| 14 | Llama 3.3 70B | 77 | #579 |
| 15 | Nova Pro | 76.6 | #569 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vellum-math · How It Works · Data refreshed daily, snapshot 2026-10-20.