Developers who want one cheap, open model for reasoning, images and coding agents should start here rather than juggling three.
Kenning's take
What it does
Mistral Small 4 merges what Mistral used to split across separate models for reasoning, image input and coding agents into one model. It has 119B total parameters with about 6B active per token, a 256k context window, and a setting to turn deeper reasoning on or off. It is released under Apache 2.0, with the weights open for download, and is available through the Mistral API.
- 119B total parameters, about 6B active per token
- Configurable reasoning effort
- Text and image input
- 256k context window
- Apache 2.0 licence
- Runs on vLLM, llama.cpp, SGLang and Transformers
Who it suits. Developers and businesses who want one efficient open model for chat, coding agents and document work.
Key facts
- Category
- AI models
- Part of
- Le Chat (Vibe by Mistral)
- Made by
- Mistral AI
- Based in
- France
- Released
- 16 March 2026
- How to get it
- through the Mistral API in Mistral Studio, and as open weights for self-hosting
- Works on
- API, self-hosted
- In the UK
- Yes
- Free plan
- open weights are free under Apache 2.0
- API price
- $0.15 per million input tokens, $0.60 per million output
FAQ
When was Mistral Small 4 released?
Mistral announced Mistral Small 4 on 16 March 2026.
Is Mistral Small 4 open source?
Yes. It is released under the Apache 2.0 licence.
How much does Mistral Small 4 cost in the API?
$0.15 per million input tokens and $0.60 per million output tokens.
Facts from the maker's own pages and the sources listed. Checked 2026-10-11.
Similar tools
Devstral 2
Mistral's open-weight coding model family, now retired from the API in favour of Mistral Medium 3.5
Mistral 3
Mistral's December 2025 family of open models, from the small Ministral 3 sizes up to Mistral Large 3
Mistral OCR 3
Mistral's December 2025 document reading model that pulls text, tables and images out of PDFs and scans
Mistral OCR 4.1
Mistral's current document reading model, returning text with layout boxes, block types and confidence scores