Version timeline
13 versions- Mistral Small 4Activemistral-small-4· 128K ctx · Fast
Unified open-weight (Apache 2.0) small model combining configurable reasoning, vision and agentic coding; MoE 119B total / ~6B active.
- Devstral 2Activedevstral-2512· 256K ctx · Reasoning
Agentic coding specialist, 72.2% SWE-bench Verified; 256K context for whole-repo sessions. Devstral Small 2 ships open-weight (modified MIT).
- Ministral 3BActiveministral-3b-2512· 128K ctx · Fast
Cheapest API option at $0.04/$0.04 per 1M; edge/on-device, open weights (Apache 2.0).
Report an issue with Mistral AI
Posts to your status feed
Pick the closest match below, edit the body, and post. Your report carries the #mistral tag automatically so it surfaces here + in the trending-tags rail. Reporting also follows Mistral AI so you’ll get status updates.
Releases are dated and source-credited where available. Catch Me Up — everything that changed across AI & smart home ↗