Mistral Medium 3.5 becomes the default in Le Chat and replaces Devstral 2 in Vibe CLI — 77.6% SWE-Bench Verified at open weights with long-horizon agentic tooling
Mistral announced this week that Medium 3.5 — the unified Magistral-reasoning + Devstral-coding open-weight model from May — becomes the default for Le Chat conversations and replaces Devstral 2 in Vibe CLI, Mistral's coding agent. Medium 3.5 scores 77.6% on SWE-Bench Verified with strong agentic capabilities for long-horizon multi-tool workflows.
The product-side commitment is the substantive read. Open-weight model releases get benchmark coverage; making the same model the default in the lab's own consumer assistant and the canonical CLI coding agent says Mistral is treating Medium 3.5 as production-quality, not research-grade. For developers comparing open weights against proprietary mid-tier — Gemini 3.5 Flash, GPT-5.5, Claude Sonnet 4.5 — the Vibe CLI default puts 77.6% SWE-Bench Verified inside the same workflow shape as Claude Code and OpenAI Codex.
The competitive frame is the open-weight category narrowing the production gap. Mistral Large 3 and Mistral Small 4 already ship under Apache 2.0 — the most permissive open-weight tier in the frontier-adjacent category. Adding Medium 3.5 as the operational default for Mistral's own consumer and developer surface signals that the company sees the open-weight tier as the destination, not the cheap alternative. Combined with this morning's extended-context patch coverage, the open-weight category gained two production-quality signals in 24 hours.
Mistral AI — Mistral AI Newsroom → · Releasebot — Mistral Release Notes - June 2026 Latest Updates → · Hugging Face — Best Open-Source LLM Models in 2026 →