POLAR-Bench: leaderboard
Metric: Overall Mean (self-reported). Source: benchmarklist.com. 22 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GLM-5.1 | 94.34 |
| 2 | GPT-5.4 | 92.33 |
| 3 | Gemma 4 31B | 91.2 |
| 4 | Gemma 4 E4B | 90.87 |
| 5 | DeepSeek V3.1 | 90.81 |
| 6 | Llama 3.3 70B | 86.23 |
| 7 | Kimi K2.5 | 85.4 |
| 8 | Gemma 3 27B | 83.16 |
| 9 | DeepSeek R1 Distill Qwen 32B | 81.19 |
| 10 | Qwen 3 32B | 80.87 |
| 11 | GLM-4.7 Flash | 80 |
| 12 | Gemma 4 E2B | 79.74 |
| 13 | Ministral 3 8B | 76.12 |
| 14 | GPT-OSS-120B | 74.62 |
| 15 | Ministral 3 14B | 71.96 |
Interactive version: theaggregate.ai/benchmark?slug=polar-bench · How It Works · Data refreshed daily, snapshot 2026-09-05.