Le Chat becomes Vibe: what Mistral announced.
On May 28, 2026, Mistral AI announced that Le Chat is now Vibe: "one agent and one licence across work and code", with every conversation, setting and plan carried over. Nothing to migrate on your side: your account continues under the new name.
It is not just a relabel. Where Le Chat was a conversational assistant, Vibe presents itself as an agent able to carry out long, multi-step tasks.
What Vibe does.
Mistral describes Vibe as a single agent with two modes:
- Work mode, on web and mobile: it catches up across your inbox and calendar, runs deep research, drafts deliverables and orchestrates recurring processes. It picks its tools and streams its progress as it goes.
- Code mode: it takes a development request all the way to a merged change, in the web app, the editor (VS Code) or the terminal. It builds features, fixes bugs, refactors code and ships reviewable pull requests.
Around the agent, you find what a modern AI assistant should offer: web search, image generation, document libraries, task scheduling, connectors and custom MCP connectors (in beta), reusable skills.
Plans and pricing.
Here are the plans shown on mistral.ai on September 27, 2026, in US dollars excluding tax:
| Plan | Price | What it includes |
|---|---|---|
| Free | $0 | Limited messages, web searches and coding sessions, image generation, 100+ connectors, $10/month in API credits |
| Pro | $14.99/month | More messages and searches, all-day coding (CLI, IDE, web), $30/month in API credits, chat and email support |
| Team | $24.99/user/month | Collaborative workspace, up to 30 GB storage per user, domain name verification, data export |
| Enterprise | Custom quote | Custom models, agents and workflows, audit logs, SAML SSO, white label, private deployments |
Two details. Verified students get Pro for $5.99 a month. And Mistral states that using your conversations for model training can be turned off.
And the Mistral API.
For developers and companies integrating Mistral into their own tools, the API is priced per million tokens, input and output counted separately. Mistral gives the example of Mistral Large at $0.5 per million input tokens and $1.5 output. Batch processing cuts the price by 50%, and caching repeated prompts cuts input cost by up to 90%.
On model choice, Mistral recommends Mistral Medium for most tasks and coding, Mistral Small for cost-sensitive projects, its OCR for documents and Voxtral for audio.
Why Mistral matters for a European business.
Mistral is a European player, and that is a concrete argument for a company handling sensitive data:
- Sovereignty. Mistral emphasizes in-region inference and European infrastructure. For customer, HR or financial data, that is a real compliance criterion.
- Private deployment. The Enterprise offer enables private deployments, and the models can be self-hosted, under a Mistral licence for commercial production use.
- Solidity. In September 2026, Mistral announced a €3 billion raise at a valuation of more than €21 billion.
The right reflex remains choosing the model by use case, not by flag. Our comparison of the main AIs for business: Which AI to Choose in 2026. And to go beyond chat, with an AI that acts in your tools: our services.
Mistral Vibe (formerly Le Chat): what people ask us.
Does Mistral's Le Chat still exist?
Yes, under the name Vibe. Mistral announced it on May 28, 2026: Le Chat becomes Vibe, one agent and one licence across work and code, with every conversation, setting and plan carried over.
Is Mistral Vibe free?
Yes, on the Free plan: limited messages, web searches and coding sessions, image generation, 100+ connectors and $10 a month in API credits to test the models. Pro raises those limits.
How much does Mistral Pro cost?
$14.99 a month excluding tax, as shown on mistral.ai. Verified students can get it for $5.99 a month. Pro includes more messages and searches, all-day coding and $30 a month in API credits.
What is the difference between Vibe and Vibe for Code?
Vibe is Mistral's single agent. Its Work mode handles long tasks (inbox, research, drafting); its Code mode and Vibe for Code handle development in the terminal, the editor or remotely.
Can I self-host Mistral's models?
Yes, according to Mistral: its models can be self-hosted. Mistral states that commercial deployments fall under a Mistral licence, with their own terms for production use. The Enterprise offer also provides private deployments.
Is Mistral a good option for a European company?
It is a European player that emphasizes sovereignty and in-region inference. For sensitive data or compliance requirements, that is a real criterion. The right choice then depends on your uses: see our comparison Which AI to Choose.