Benchmarks
European languages
Calidad del modelo en benchmarks traducidos a idiomas europeos. Las filas EU21 provienen de la evaluación OpenGPT-X publicada con Teuken 7B Instruct v0.6, que traduce ARC, HellaSwag, TruthfulQA y MMLU a los 21 idiomas de la UE que cubre e informa un promedio por modelo. La fila MMMLU es la cifra MMLU multilingüe propia de Google para Gemma 4.
The EU21 numbers are OpenGPT-X’s, taken from the Teuken 7B Instruct v0.6 model card and the accompanying evaluation preprint; they are machine-translated benchmarks, not native-language test sets. The MMMLU number is Google’s own. LLM EU has run no evaluation and has not verified either harness.
| Modelos | score | source |
|---|---|---|
| Teuken 7B Instruct v0.6 | 0.57 | link |
| Llama 3.1 8B Instruct | 0.563 | link |
| Mistral 7B Instruct v0.3 | 0.527 | link |
| Aya 23 8B | 0.485 | link |
| Pharia-1-LLM-7B-control-aligned | 0.417 | link |
| Gemma 4 31B IT | 88.4 | link |