Commonsense Reasoning on Hellaswag 24 official EU languages
84.3AccuracyMistral-3.2-24B
Evaluation Results
| Method | Links | |
|---|---|---|
| Mistral-3.2-24BWeights accessibility=Open-weights, Regional origin=European2026.02 | 84.3 | |
| Qwen-3-32BWeights accessibility=Open-weights, Regional origin=Non-European2026.02 | 80.5 | |
| Qwen-3-30B-A3BWeights accessibility=Open-weights, Regional origin=Non-European2026.02 | 79.3 | |
| Qwen-3-14BWeights accessibility=Open-weights, Regional origin=Non-European2026.02 | 77.5 | |
| Gemma-3-27BWeights accessibility=Open-weights, Regional origin=Non-European2026.02 | 76.4 | |
| Llama-3.3-70BWeights accessibility=Open-weights, Regional origin=Non-European2026.02 | 74.7 | |
| Gemma-3-12BWeights accessibility=Open-weights, Regional origin=Non-European2026.02 | 74.5 | |
| Apertus-70BWeights accessibility=Fully-open, Regional origin=European2026.02 | 68.6 | |
| EuroLLM-22BWeights accessibility=Fully-open, Regional origin=European2026.02 | 62.6 | |
| Apertus-8BWeights accessibility=Fully-open, Regional origin=European2026.02 | 50.9 | |
| EuroLLM-9BWeights accessibility=Fully-open, Regional origin=European2026.02 | 49.9 | |
| OLMo-3.1-32BWeights accessibility=Fully-open, Regional origin=Non-European2026.02 | 49.2 | |
| Llama-3.1-8BWeights accessibility=Open-weights, Regional origin=Non-European2026.02 | 37.7 | |
| OLMo-3-7BWeights accessibility=Fully-open, Regional origin=Non-European2026.02 | 30 |