Ministral 3 8B Instruct 2512
Mistral AI · 2025-12-02 · 8.9B parameters
Ministral 3 8B Instruct 2512 is Mistral AI's December 2025 instruction-post-trained, non-reasoning checkpoint for edge and local deployment. The open-weight Apache-2.0 model combines an 8.4B-class dense language model with a frozen 410M-parameter vision encoder and a trained multimodal projector, accepts text and image input, produces text, and supports a 262,144-token context; this repository is the FP8 checkpoint, while Base and Reasoning are separate releases.
Benchmark scores
| Benchmark | Score |
|---|---|
| LisanBench Difficulty-Weighted | 11.8 |
| LisanBench Path Length | 32.3 |