Mistral Small 3.1
Mistral Small 3.1 adds vision and 128K context to the 24B Apache 2.0 model and, per Mistral, beats Gemma 3 and GPT-4o Mini.
Same 24B size, now multimodal with 128K context and about 150 tokens/s. Targets document verification, visual inspection and local use.
- Date
- Monday, 17 March 2025
- Lab
- Mistral AI
- Kind
- open-weights
- Access
- open weights
Figures
| Measure | Value | Measured by |
|---|---|---|
| Context window | 128K tokens 24B, Apache 2.0 | company |
Sources
This record was checked against its sources on 6 October 2026. How we check