Description
The smallest model in the Ministral 3 family, combining a 3.4B language model with a 0.4B vision encoder for efficient edge deployment.
Stats
16.6K Downloads
3 stars
Capabilities
Vision Input
Trained for tool use
Minimum system memory
2GB
Tags
3B
mistral3
Last updated
Updated 3 days agobyREADME
Ministral 3 3B by mistralai
The smallest model in the Ministral 3 family, combining a 3.4B language model with a 0.4B vision encoder for efficient edge deployment.
Supports context length of 256k tokens.
Vision-enabled for image analysis and multimodal tasks.
Multilingual support across dozens of languages including English, French, German, Spanish, Italian, Chinese, Japanese, Korean, Portuguese, and more.
Native function calling and JSON output generation.
Apache 2.0 License
Sources
The underlying model files this model uses
Based on