Mistral Medium 3.5
Language modelOpen weightsCard summarizedFrontier-class multimodal model optimized for agentic and coding use cases
Active● Re-checked 12 Oct 2026 · 27 facts · 0 corrections · 1 unconfirmedSource: Mistral Medium 3.5 model card (Hugging Face) and launch post
Use it via API
Where to call it
- Mistral API
mistral-medium-3-5 - Self-host (weights)
mistralai/Mistral-Medium-3.5-128B
Good for
- Long-horizon coding and productivity work
- Agentic tasks: reliable multi-tool calls and structured (JSON) output
- Vision and multilingual chat
Overview
Medium 3.5
Frontier-class multimodal model optimized for agentic and coding use cases
- Cost
- $1.50 in
- Context
- 256K
- Input / output
- $1.50 / $7.50
- Max output
- —
- Knows up to
- —
- Released
- 28 Apr 2026
Mistral's first flagship merged model: a dense 128B model with a 256K context window that combines instruction following, reasoning and coding in one set of weights. Reasoning effort is set per request. Released as open weights under a Modified MIT license. HF model card
mistral-medium-3-5- Input / output per 1M
- $1.50 / $7.50
- Context window
- 256K
- Max output
- —
- Knows up to
- —
- Reasoning
- Configurable (reasoning_effort: none or high)
- Weights
- Open · Modified MIT
About this card. A model card is the maker's own account of a model: what it's for, how it was trained and tested, and where it falls short. This page condenses Mistral AI's Mistral Medium 3.5 model card (Hugging Face) and launch post. Every claim carries its source page.