INCLUDE-base-44 European Languages: leaderboard
European-language slice of INCLUDE-base-44, evaluating multilingual LLMs on knowledge- and reasoning-centric multiple-choice questions across 20 European languages.
Metric: Average Accuracy (%). Source: huggingface.co. Source: huggingface.co. Saturation forecast: Around January 2027. 35 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GaMS3-12B-Instruct | 65.79 | #704 |
| 2 | Qwen 3 14B | 65.07 | #537 |
| 3 | cogito-v1-preview-qwen-14B | 64.36 | |
| 4 | Gemma 3 12B (IT) | 63.93 | #664 |
| 5 | Qwen 2.5 14B Instruct | 61.66 | #647 |
| 6 | Qwen 2.5 14B | 61.09 | #725 |
| 7 | Qwen 3 8B | 61.05 | #674 |
| 8 | Phi-4 | 58.82 | #706 |
| 9 | Apertus-8B-Instruct-2509 | 57.89 | #1061 |
| 10 | Llama 3.1 8B Instruct | 55.34 | #1035 |
| 11 | EuroLLM-9B-Instruct | 55.1 | #1054 |
| 12 | Qwen 2.5 7B | 54.68 | #922 |
| 13 | Qwen 2.5 7B Instruct | 54.36 | #862 |
| 14 | Mistral Nemo Instruct (2407) | 53.2 | #936 |
| 15 | Bielik-11B-v2.5-Instruct | 51.36 | #567 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=include-base-44-european-languages · How It Works · Data refreshed daily, snapshot 2026-10-20.