Le Chat is Mistral AI's assistant — a Paris-built rival to ChatGPT and Claude that answers questions, writes and runs Python, generates images, and now plugs into Gmail, Slack, GitHub and Notion to carry out multi-step work. In May 2026 Mistral renamed it Vibe and split it into three modes: Chat for quick answers, Work for agent tasks across your inbox and files, and Code for building software. It runs on the web, on iPhone and Android, and inside VS Code, and it's aimed at anyone who wants a serious assistant with a free tier that hands over frontier models, a real code sandbox and a stack of app connectors without asking for a card.
Key Features
- A free plan that isn't hollowed out — frontier chat models, image generation, a working Python code interpreter and dozens of app connectors, no credit card, held back only by a soft cap of roughly 25 messages a day.
- The code interpreter runs genuine Python in a sandbox with pandas, numpy and matplotlib already installed, so you can drop in a messy CSV and get back cleaned data and charts instead of snippets to paste somewhere else.
- Work Mode links Gmail, Outlook, Slack, SharePoint and GitHub, lays out its plan, and waits for a thumbs-up before it touches anything; scheduled tasks can repeat the same job daily, weekly or monthly.
- Deep Research builds structured, source-cited reports; voice input uses Mistral's own Voxtral model; Projects group related chats with their own library, and opt-in Memories carry context between conversations.
- European by default — Mistral is based in France and runs on EU infrastructure, which is the main reason regulated teams choose it over the American assistants.
Screenshots
Why You'll Love Le Chat
The free tier gives away what rivals charge for
On the $0 plan you get Mistral's current frontier models for chat, image generation through Black Forest Labs' FLUX, the Python code interpreter, and a large catalogue of connectors — Mistral's marketing now claims more than 100, and the public connector directory listed around 65 in mid-2026. ChatGPT and Claude gate code execution, integrations and their best models behind $20 subscriptions. The catch is volume: Mistral soft-caps the free plan near 25 messages a day, and image generation and Deep Research burn through that budget faster than plain text, so a busy afternoon can lock you out until the next day. For light daily use — a few questions, the odd chart, an image here and there — it's genuinely enough.
What the paid plans cost and unlock
Pro is $14.99 a month, with verified students paying $5.99. That undercuts the comparable ChatGPT Plus, Claude Pro and Google AI Pro tiers, which all sit at $20. Pro lifts the daily cap to roughly 150 messages, turns on extended reasoning and the Mistral Medium 3.5–powered Work and Code agents, adds Flash Answers, and bundles a small monthly API credit for Mistral Studio (Mistral lists $15, or $30 for students; the Free plan lists $10). Team is $24.99 per user per month, billed for at least two seats, and adds 30 GB of storage per person, domain verification, data export, higher Flash Answer limits and a shared workspace. Enterprise is custom and brings SAML SSO, audit logs, custom models and private or white-label deployment. One thing to keep straight: a Le Chat subscription does not cover API usage — that's billed separately, beyond the small included credit.
Work Mode is where it's actually headed
Since the Vibe rebrand the product leans less on chatting and more on doing: reading your inbox, pulling figures out of a spreadsheet, drafting a report and filing it in Notion or SharePoint. It shows its plan first and you approve each step, so it's not firing off actions blind. That's genuinely useful if your work already lives in Google Workspace, Outlook, Slack and GitHub. If you just want a fast chat box, the three-mode layout is more machinery than you need, and the plain Chat mode is the part to stick with.
Category: AI Assistant
Platform: Web, iOS, Android (plus a VS Code extension and CLI for Code mode)
Price: Free with a ~25 messages/day soft cap; Pro $14.99/month ($5.99 students); Team $24.99/user/month; Enterprise custom
Frequently Asked Questions
Yes. The free plan needs no card and includes frontier chat models, image generation, the Python code interpreter and app connectors, soft-capped at around 25 messages a day — image generation and Deep Research use that budget up faster than text. Pro is $14.99/month (about $5.99 for students) for roughly 150 messages a day plus the Work and Code agents; Team is $24.99/user/month; Enterprise is custom.
No. Every response runs on Mistral's servers in Europe, and the code interpreter sandbox is cloud-side too, so you need a connection. The iOS and Android apps just talk to the same backend — there's no on-device model and nothing works in airplane mode.
On a phone, not much by default: it asks for the microphone only when you start voice mode, and for camera or photo access only when you attach or generate images. Notifications are optional. The bigger consent step is connectors — linking Gmail, Google Drive, Slack or GitHub runs an OAuth flow where you grant specific scopes, and you can revoke any of them later from settings.