Open-R1 Eval Leaderboard: leaderboard
Open-R1 released-model reasoning leaderboard aggregating LightEval results across math, GPQA, AIME, MATH-500, and related reasoning tasks.
Metric: Average Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 37 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 32B | 73.74 | #443 |
| 2 | Qwen 3 30B A3B | 72.94 | #493 |
| 3 | Qwen 3 14B | 72.67 | #537 |
| 4 | QwQ-32B | 70.58 | #421 |
| 5 | Qwen 3 8B | 69.79 | #674 |
| 6 | DeepSeek R1 Distill Qwen 32B | 65.96 | #655 |
| 7 | DeepSeek R1 Distill Llama 70B | 65.75 | #594 |
| 8 | Qwen 3 4B | 65.61 | #829 |
| 9 | DeepSeek R1 Distill Qwen 14B | 63.12 | #844 |
| 10 | DeepSeek-R1-Distill-Qwen-7B | 51.57 | #1415 |
| 11 | QwQ 32B-Preview | 51.01 | #651 |
| 12 | Qwen 3 1.7B | 48.95 | #1196 |
| 13 | DeepSeek R1 Distill Llama 8B | 47.77 | #1291 |
| 14 | DeepSeek R1 Distill Qwen 1.5B | 34.46 | #1551 |
| 15 | Qwen 2.5 Coder 32B Instruct | 28.73 | #528 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=open-r1-eval-leaderboard · How It Works · Data refreshed daily, snapshot 2026-10-20.