U-MATH - Integral Calculus — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O1 | 83.17 |
| 2 | O3 Mini | 82.21 |
| 3 | Gemini 2.0 Flash (01-21) (Thinking) | 82.21 |
| 4 | DeepSeek R1 | 76.92 |
| 5 | O1 Mini | 64.9 |
| 6 | QwQ 32B-Preview | 62.5 |
| 7 | QVQ-72B-Preview | 45.19 |
| 8 | DeepSeek V3 | 38.94 |
| 9 | Qwen2.5-Math-72B-Instruct | 38.94 |
| 10 | Qwen 2.5 32B Instruct | 32.21 |
| 11 | Athene-V2-Chat | 31.25 |
| 12 | Gemini 1.5 Flash | 30.77 |
| 13 | Mistral Large 2 (Nov) Instruct (2411) | 30.77 |
| 14 | Qwen 2.5 72B Instruct | 30.29 |
| 15 | Phi-4 | 28.37 |
Interactive version: aibenchmarks.dev/benchmark?slug=u-math-integral-calculus · How the rankings work · Data refreshed daily, snapshot 2026-07-08.