Model
Inference & Serving
Ministral
1Mistral's 2compact 3models 4for 5edge 6computing
the six Ws · specification
W1
Who
Mistral AI released the Ministral 3B and 8B models.
W2
What
Small, efficient language models optimized for on-device and edge deployment with low latency.
W3
Where
Distributed on Hugging Face and Mistral's model hub.
W4
When
Ministral models were released in October 2024.
W5
Why
They targeted local, privacy-sensitive, and low-resource deployments, competing with Gemma and Phi in the small-model space.
W6
With
Runs via transformers, vLLM, or llama.cpp, and supports quantized deployment on edge hardware.
W7
Watch
Released by Mistral AI; Ministral 8B weights use the Mistral Research License (non-commercial), while Ministral 3B is more permissively licensed for research.
alternatives
works with