Omni-MATH: leaderboard
Olympiad-level mathematics benchmark spanning domains and difficulty bands, using structured grading to evaluate advanced mathematical reasoning.
Metric: Overall Accuracy (%). Source: omni-math.github.io. Status: saturated. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O1 Mini | 60.54 |
| 2 | O1 Preview | 52.55 |
| 3 | Qwen2.5-Math-72B-Instruct | 36.2 |
| 4 | Qwen2.5-Math-7B-Instruct | 33.22 |
| 5 | GPT-4o | 30.49 |
| 6 | Claude 3.5 Sonnet | 26.23 |
| 7 | DeepSeek Coder V2 | 25.78 |
| 8 | Llama 3.1 70B Instruct | 24.16 |
Interactive version: theaggregate.ai/benchmark?slug=omni-math · How It Works · Data refreshed daily, snapshot 2026-09-05.