LiveMathBench 202412: leaderboard
Models answer 119 recent competition and entrance-exam math questions (CNMO, Chinese gaokao, AMC, Putnam), each in English and Chinese, 48 sampled answers per question; the score is mG-Pass@16, which rewards getting a question right consistently across draws rather than once. Qwen2.5-72B-Instruct checks each answer against the reference and is itself withheld. Unlike a pass@1 math set, a model that is right only on some draws scores low.
Metric: mG-Pass@16 on 238 questions (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 27 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek R1 | 77.6 | #487 |
| 2 | O3 Mini | 76.8 | #547 |
| 3 | DeepSeek R1 Distill Qwen 32B | 69.7 | #1806 |
| 4 | DeepSeek R1 Distill Llama 70B | 69.7 | #1669 |
| 5 | DeepSeek R1 Distill Qwen 14B | 68 | #2683 |
| 6 | O1 Mini | 64.8 | #776 |
| 7 | DeepSeek-R1-Distill-Qwen-7B | 63.1 | #7957 |
| 8 | QwQ 32B-Preview | 61.2 | #1628 |
| 9 | DeepSeek R1 Distill Llama 8B | 52.2 | #6937 |
| 10 | DeepSeek V3 | 47.9 | #734 |
| 11 | DeepSeek R1 Distill Qwen 1.5B | 45.1 | #9148 |
| 12 | Qwen2.5-Math-72B-Instruct | 42.3 | #1757 |
| 13 | Qwen 2.5 Max | 42.2 | #966 |
| 14 | Qwen 2.5 32B Instruct | 38.1 | #1257 |
| 15 | Qwen2.5-Math-7B-Instruct | 36.6 | #8400 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=livemathbench-202412 · How It Works · Data refreshed daily, snapshot 2026-11-17.