U-MATH - Multivariable Calculus — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O1 | 87.08 |
| 2 | Gemini 2.0 Flash (01-21) (Thinking) | 83.71 |
| 3 | DeepSeek R1 | 79.78 |
| 4 | O3 Mini | 79.78 |
| 5 | QwQ 32B-Preview | 73.03 |
| 6 | O1 Mini | 73.03 |
| 7 | DeepSeek V3 | 65.17 |
| 8 | QVQ-72B-Preview | 62.92 |
| 9 | Qwen2.5-Math-72B-Instruct | 61.8 |
| 10 | Gemini 1.5 Pro | 60.11 |
| 11 | Athene-V2-Chat | 55.62 |
| 12 | Gemini 1.5 Flash | 55.06 |
| 13 | Phi-4 | 50.56 |
| 14 | GPT-4o (2024-08-06) | 48.31 |
| 15 | Qwen 2.5 72B Instruct | 46.07 |
Interactive version: aibenchmarks.dev/benchmark?slug=u-math-multivariable-calculus · How the rankings work · Data refreshed daily, snapshot 2026-07-08.