Open PL LLM - RAG: leaderboard
Open PL LLM aggregate over retrieval-augmented Polish QA tasks.
Metric: Average RAG Score (%). Source: huggingface.co. Saturation forecast: Around October 2026. 287 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Bielik-11B-v2.2-Instruct | 76.92 | #660 |
| 2 | Bielik-11B-v2.1-Instruct | 76.79 | #673 |
| 3 | Bielik-11B-v2.3-Instruct | 76.2 | #693 |
| 4 | Mistral Large 2 (Jul) | 75.78 | #484 |
| 5 | Bielik-11B-v2.0-Instruct | 75.5 | #742 |
| 6 | Qwen 2 72B | 75.48 | #521 |
| 7 | cogito-v1-preview-qwen-14B | 75.19 | |
| 8 | Mistral Large 2 (Nov) Instruct (2411) | 74.96 | #465 |
| 9 | Llama 3.3 70B Instruct | 74.62 | #534 |
| 10 | Llama 3.1 405B Instruct FP8 | 74.34 | #389 |
| 11 | Gemma 3 27B (IT) | 73.92 | #523 |
| 12 | Llama 4 Scout Instruct | 73.89 | #539 |
| 13 | QwQ 32B-Preview | 73.8 | #651 |
| 14 | Qwen 3 14B | 73.73 | #537 |
| 15 | Llama 3.1 70B Instruct | 73.67 | #561 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=open-pl-llm-rag · How It Works · Data refreshed daily, snapshot 2026-10-20.