MT-Bench PL - Math: leaderboard
The "Math" subtask of MT-Bench PL. The Polish-language version of MT-Bench, scoring multi-turn conversational and instruction-following ability via LLM-as-judge.
Metric: Judge Score (0-10). Source: huggingface.co. Saturation forecast: Estimated already saturated. 50 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 3 27B (IT) | 8.25 | #523 |
| 2 | Qwen 2.5 14B Instruct | 8.1 | #647 |
| 3 | Mistral Small 3.1 | 7.85 | #610 |
| 4 | Mistral Small 3 | 7.83 | #627 |
| 5 | Gemma 2 27B (IT) | 7.8 | #645 |
| 6 | Mistral Large 2 (Jul) | 7.8 | #484 |
| 7 | Phi-4 | 7.7 | #706 |
| 8 | Bielik-11B-v2.3-Instruct | 7.7 | #693 |
| 9 | Qwen 2.5 32B Instruct | 7.6 | #498 |
| 10 | Gemma 3 12B (IT) | 7.45 | #664 |
| 11 | Gemma 3 4B (IT) | 7.4 | #992 |
| 12 | Mistral-Small-Instruct-2409 | 7 | #702 |
| 13 | Mixtral 8x22B | 6.9 | #766 |
| 14 | GPT-3.5 Turbo | 6.85 | #865 |
| 15 | Mistral Nemo Instruct (2407) | 6.7 | #936 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mt-bench-pl-math · How It Works · Data refreshed daily, snapshot 2026-10-20.