EuroEval Polish NLU - PoQuAD: leaderboard
The "PoQuAD" subtask of EuroEval Polish NLU. Polish-language EuroEval natural-language-understanding suite, separating NLU task performance from the broader all-task EuroEval aggregate.
Metric: Character F1 (%). Source: euroeval.com. Saturation forecast: Not forecast. 234 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen3.5-4B-Base | 65.76 | #798 |
| 2 | Llama 3.1 70B | 63.87 | #592 |
| 3 | SOLAR-10.7B-v1.0 | 63.61 | #1253 |
| 4 | gemma-3-27B-pt | 61.94 | #502 |
| 5 | Llama 3.1 8B | 61.78 | #1151 |
| 6 | gemma-3-12B-pt | 60.97 | #686 |
| 7 | Qwen 3 32B (Non-reasoning) | 60.13 | #443 (Qwen 3 32B) |
| 8 | Qwen 3 14B (Non-reasoning) | 60.13 | #537 (Qwen 3 14B) |
| 9 | EuroLLM-9B | 60.06 | #1254 |
| 10 | Bielik-11B-v2.3-Instruct | 59.91 | #693 |
| 11 | Olmo-3.1-32B-Instruct-SFT | 59.18 | #879 |
| 12 | occiglot-7B-eu5-instruct | 59.02 | |
| 13 | Gemma 4 31B | 58.93 | #200 |
| 14 | Mistral Small 3.2 | 58.53 | #605 |
| 15 | Ministral-3-14B-Reasoning-2512 | 57.94 | #540 |
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-poquad · How It Works · Data refreshed daily, snapshot 2026-10-20.