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. Status: saturated. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Kimi K2.5 | 98 |
| 2 | O3 Mini | 97.9 |
| 3 | Claude Sonnet 4.6 | 97.8 |
| 4 | Claude Opus 4.6 | 97.6 |
| 5 | DeepSeek R1 | 97.3 |
| 6 | O1 | 96.4 |
| 7 | Claude 3.7 Sonnet (Thinking) | 96.2 |
| 8 | DeepSeek V3 (0324) | 94 |
| 9 | O1 Mini | 90 |
| 10 | Gemini 2.0 Flash | 89.7 |
| 11 | Gemma 3 27B | 89 |
| 12 | Claude 3.7 Sonnet | 82.2 |
| 13 | Claude 3.5 Sonnet | 78 |
| 14 | Llama 3.3 70B | 77 |
| 15 | Nova Pro | 76.6 |
Interactive version: theaggregate.ai/benchmark?slug=vellum-math · How the rankings work · Data refreshed daily, snapshot 2026-07-22.