Ministral 3 8B 2512

An 8-billion parameter language model from Mistral with vision capabilities and a 262K token context window.

Strengths

Handles both text and image inputs while maintaining efficient performance across a wide range of tasks.

Best for

Applications requiring multimodal processing where model size and latency matter, positioned between the smaller Ministral 3B and larger 14B variant.

Limitations

Less capable than larger models in the Mistral lineup for complex reasoning or generation tasks; released as part of the December 2025 Ministral 3 family.