On June 1, Mistral did the most on-the-nose thing in this entire category: it renamed its Le Chat assistant to Vibe and relaunched it as a two-headed agent platform. There’s Vibe for Work, a productivity agent that plugs into Google Workspace, Outlook, SharePoint, Slack, and GitHub to summarize email, analyze spreadsheets, draft reports, and run recurring tasks — and Vibe for Code, a coding agent that lives in a web interface, a new VS Code extension, and the existing Mistral CLI, and will build features, fix bugs, refactor, and open pull requests. The name is a marketing land-grab on “vibe coding”; the product underneath is Mistral’s most serious push yet into the agentic market that OpenAI, Anthropic, and Google currently own.

What’s actually new here

The rebrand matters less than the architecture. Vibe for Code ships remote agents powered by Mistral Medium 3.5: coding sessions that run in Mistral’s cloud rather than on your laptop, spawnable from the CLI or the Vibe web app. You can start a session locally and “teleport” it to the cloud to keep running while you close your machine — the same async-agent pattern Anthropic shipped as Routines and GitHub shipped as Copilot Workspace. Slack integration for Code Mode is slated to arrive later in June, so you’ll be able to kick off and monitor coding runs from a channel.

If that sounds familiar, it should. Mistral isn’t inventing a new interaction model; it’s matching the table stakes — chat agent, IDE extension, CLI, cloud-run async sessions — that the leaders set over the past two months. The differentiator isn’t the feature list.

The real pitch: a European home for your agent

Mistral’s wedge is data residency and independence. For a French or EU company that’s nervous about routing internal documents and source code through a U.S. lab, Vibe is the credible homegrown option — one agent that touches both the office stack and the codebase, hosted under European rules. That’s a genuine reason to pick it, and it’s the reason most likely to win deals. It has very little to do with whether the coding model is the best on the market (it isn’t — Mistral Medium 3.5 trails the frontier Claude and GPT models on the hard agentic benchmarks).

Pricing

There’s a free tier for light use, Pro at $14.99/mo, Teams at $24.99 per user/mo, and custom Enterprise pricing. There’s also a $5.99/mo student plan, and API access at $1.50 per million input tokens and $7.50 per million output. The Pro price undercuts the $20 anchor most competitors sit at, which is on-brand for Mistral — cheaper, open-leaning, European.

Who it’s for

EU-based teams already committed to the Mistral stack, or anyone whose procurement team treats U.S. data routing as a blocker. The “one agent for work and code” framing is also genuinely useful for small operator-founder teams who don’t want a separate tool for inbox triage and a separate one for shipping features.

Who it’s not for

If you’re a non-technical founder looking for Lovable-style “describe an app, watch it build,” this isn’t that — Vibe for Code assumes a real development workflow (a repo, a VS Code or CLI habit, PR review). And if you’re optimizing purely for coding-agent quality regardless of geography, the frontier still lives elsewhere.

Verdict

A smart, well-timed repositioning rather than a breakthrough. Vibe’s strength is consolidation and sovereignty, not raw capability: one European agent spanning office work and code, priced below the field. For Mistral-stack and EU-compliance-driven teams it’s an easy tool to put on the shortlist. Everyone else should weigh it against the better-resourced coding agents before switching — and watch whether Mistral Medium 3.5 closes the quality gap or just the price one.

Was this helpful?