Polish Knowledge Benchmark - Art: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 68 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.6 Flash | 94.39 |
| 2 | Gemini 3 Flash (Preview) | 94.12 |
| 3 | Gemini 3.1 Flash Lite | 91.44 |
| 4 | Gemini 3.1 Flash Lite (Preview) | 90.91 |
| 5 | DeepSeek V4 Flash | 89.04 |
| 6 | Gemini 3.5 Flash Lite | 88.5 |
| 7 | Gemini 2.5 Flash | 87.97 |
| 8 | Grok 4.1 Fast | 87.97 |
| 9 | DeepSeek V3.2 | 87.7 |
| 10 | DeepSeek V3 (0324) | 86.9 |
| 11 | Qwen 3.5 397B A17B | 86.36 |
| 12 | Mistral Large 3 | 86.36 |
| 13 | Gemini 2.0 Flash (001) | 86.36 |
| 14 | Mistral Medium 3.1 | 85.03 |
| 15 | Gemma 4 31B (IT) | 83.16 |
Interactive version: theaggregate.ai/benchmark?slug=polish-knowledge-benchmark-art · How It Works · Data refreshed daily, snapshot 2026-09-19.