WildBench v2 - release-reported score
Chat & Writing
WildBench v2 score as printed by a model release that does not state the judge configuration. Kept as a diagnostic. Higher is better.
Top models (higher is better)
| Model | Score |
|---|---|
| Mistral Small 3.2 24B Instruct 2506 | 65.3 |
| Mistral Small 3.1 24B Instruct 2503 | 55.6 |