EuroEval Polish NLU: leaderboard
Polish-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: NLU Average Score (%). Source: euroeval.com. Saturation forecast: Estimated already saturated. 241 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 4 31B (IT) | 68.96 | #176 |
| 2 | Llama 3.1 70B | 67.76 | #592 |
| 3 | Gemini 3 Pro (Preview) | 67.69 | #66 |
| 4 | GPT-5.4 Mini (2026-03-17) (High) | 67.34 | #169 (GPT-5.4 Mini (2026-03-17)) |
| 5 | Bielik-11B-v2.3-Instruct | 66.58 | #693 |
| 6 | Gemini 3 Flash (Preview) (Non-reasoning) | 66.57 | #82 (Gemini 3 Flash (Preview)) |
| 7 | Gemma 4 26B A4B (IT) | 66.36 | #297 |
| 8 | GPT-5 | 66.28 | #98 |
| 9 | gemma-3-27B-pt | 66.17 | #502 |
| 10 | Gemini 3.1 Flash Lite (Preview) | 65.69 | #194 |
| 11 | Claude Sonnet 4.5 (Thinking) | 65 | #142 (Claude Sonnet 4.5) |
| 12 | Gemini 3 Flash (Preview) (Thinking) | 64.8 | #82 (Gemini 3 Flash (Preview)) |
| 13 | Claude Sonnet 4.6 | 64.79 | #90 |
| 14 | Mistral Small 3.2 | 64.51 | #605 |
| 15 | Qwen 3.5 35B A3B Base | 64.48 | #310 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=euroeval-polish-nlu · How It Works · Data refreshed daily, snapshot 2026-10-20.