JudgeBench Math — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 52 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.6 Plus | 96.43 |
| 2 | Qwen 3.5 397B A17B | 94.64 |
| 3 | Qwen 3.6 35B A3B | 94.64 |
| 4 | Qwen 3.5 9B | 93.75 |
| 5 | Kimi K2.5 | 93.75 |
| 6 | Qwen 3.5 4B | 93.75 |
| 7 | GPT-5 Mini | 92.86 |
| 8 | MiniMax-M2.5 | 92.86 |
| 9 | Qwen 3.5 35B A3B | 92.86 |
| 10 | MiMo-V2-Flash | 92.86 |
| 11 | GLM-5 FP8 | 92.86 |
| 12 | Gemini 3.1 Pro (Preview) | 91.96 |
| 13 | Qwen 3.5 122B A10B | 91.96 |
| 14 | DeepSeek V3.2 Speciale | 91.96 |
| 15 | GPT-5.5 | 91.07 |
Interactive version: theaggregate.ai/benchmark?slug=judgebench-math · How the rankings work · Data refreshed daily, snapshot 2026-07-22.