Mistral AI

Le Chat

Provider

Mistral AI

Price

$0-24.99/mo

Platforms

Web, iOS, Android

Launched

Feb 2024

Updated June 2026

Le Chat is the consumer AI assistant from Mistral AI, the Paris-based company widely seen as Europe’s leading independent AI lab. As of 28 May 2026, Le Chat has been rebranded as Vibe — “one agent and one licence across work and code” — but it is the same product, and existing accounts kept every conversation, setting and plan. If you searched for “Le Chat”, you are now looking at Vibe. Its distinguishing pitch is unchanged: data hosted in the European Union by default, strong multilingual performance, very fast responses, and an affordable Pro plan at $14.99/month.

Le Chat (Vibe) is best for people who want a privacy-first, EU-built alternative to American assistants. It is not the model to beat Claude or ChatGPT on independent coding benchmarks, but it is competitive on everyday tasks, faster than most rivals, cheaper at the Pro tier, and built natively under GDPR. This guide covers what the app does today, what changed in the Vibe rebrand, and who should use it.

Quick stats

Pricing breakdown

Le Chat (Vibe) has four tiers: Free, Pro, Team and Enterprise. Mistral lists consumer prices in US dollars on its global pricing page; EU customers are billed in euros at the same figures (Pro is €14.99). All prices below are in USD.

Free — $0

The free tier is genuinely capable and unusually generous on features for a no-cost plan. You get:

• Access to current Mistral models — including the models that power Vibe’s Work Mode
• Web search and Flash Answers — Cerebras-accelerated responses at roughly 1,000+ words per second
• Image generation and editing — powered by Black Forest Labs’ FLUX
• 40+ connectors — link tools such as GitHub, Google Drive, SharePoint and more
• Memories (up to 500) — persistent context across conversations (beta)
• Voice input — powered by Mistral’s open-source Voxtral speech model
• EU data hosting by default

Limitations: lower message, web-search and deep-research limits than Pro; less coding capacity in Code Mode; no no-telemetry mode (free prompts may be used to improve the service unless you opt out).

Worth it for: casual users, students, and anyone wanting a fast, private, European alternative to ChatGPT’s free tier.

Pro — $14.99/month

The core paid plan, and one of the cheapest frontier-assistant subscriptions on the market. Everything in Free, plus:

• Higher limits — far more messages, web searches and deep-research reports
• Expanded coding capacity — all-day coding through Vibe’s Code Mode
• Deeper reasoning — more access to Mistral’s reasoning models
• No-telemetry mode — stops Mistral using your prompts and outputs for training
• Larger Memories allowance

Student price: verified students pay $5.99/month — among the lowest student AI plans available.

Worth it for: regular users who want frontier-class features at roughly a quarter of the cost of ChatGPT Plus or Claude Pro at $20/month.

Team — $24.99/user/month

For small teams and organisations. Everything in Pro, plus:

• Shared workspace with admin controls
• Data excluded from training by default — no opt-out required
• More storage, data export and domain verification
• Annual billing — $19.99/user/month

Independent coverage notes that Team’s value sits in administration — storage, exports and domain controls — rather than a higher usage ceiling than Pro (WinBuzzer).

Worth it for: small teams that need shared workspaces, admin governance and a no-training-by-default guarantee.

Enterprise — Custom pricing

For large organisations with strict requirements:

• On-premises and private-cloud deployment — full data residency
• Custom connectors and model customisation via Forge
• Admin-level permissions and governance
• Dedicated support and solutioning

Le Chat Enterprise launched in May 2025 with native connectors for SharePoint, Google Drive, OneDrive, Gmail and Google Calendar, and has expanded since. Enterprise customers include ASML, BNP Paribas, CMA CGM, HSBC and BMW.

Models powering Le Chat

Vibe runs on Mistral’s own models, “optimised for reasoning, agentic tasks, tool calls and coding” (Mistral). You don’t pick a single model the way you do in ChatGPT; the app routes to the right Mistral model for the task. For full specs and pricing, see the Mistral models page.

Mistral Medium 3.5

The flagship that powers most of Vibe’s work — a reasoning and agentic model announced alongside the Vibe remote agents launch. It handles long-horizon Work Mode tasks, tool calls and analysis.

Mistral Large 3

Mistral’s largest open-weight model: a Mixture-of-Experts design with 41 billion active parameters and 675 billion total, released under the Apache 2.0 licence. Independent testing placed it among the leading open-weight non-reasoning models on the LMArena leaderboard, though it sits below closed frontier models such as Claude Opus 4.8 and GPT-5.5 on the hardest benchmarks.

Magistral

Mistral’s dedicated reasoning family, built for multi-step logic with transparent chain-of-thought. Magistral ships in a Medium (enterprise) variant and an open-source 24B Small variant.

Devstral 2

The agentic coding model behind Code Mode. It builds on Devstral Small’s 46.8% on SWE-bench Verified — a respectable score for an efficient, open model, but below the 80%+ figures posted by the frontier coding leaders.

Mistral Small 4, Pixtral and Voxtral

Mistral Small 4 unifies reasoning (Magistral), vision (Pixtral) and coding (Devstral) into one model with a reasoning_effort control. Pixtral handles image understanding; Voxtral powers voice input and was open-sourced in July 2025. Image generation and editing are not a Mistral model — they run on Black Forest Labs’ FLUX.

Features deep dive

Work Mode — the agent (web and mobile)

Work Mode is the centre of the Vibe rebrand. It is an agent for complex, multi-stage tasks: it maps out a plan, gets your sign-off, then works across your connected apps to finish the job. Today it can:

• Search enterprise knowledge across Google Workspace, Outlook, SharePoint, Slack, GitHub and custom connectors
• Analyse structured data — connect a database or upload a spreadsheet and Vibe surfaces patterns and renders charts in the conversation
• Synthesise documents in the Canvas tool — briefs, reports, RFP responses, board decks — ready to push to Notion, SharePoint or your inbox
• Schedule tasks to run once or on a daily, weekly or monthly cadence
• Run reusable skills that turn repeatable workflows into saved automations

Every step is visible, and each tool call and reasoning chain can be expanded to its inputs and outputs.

Code Mode and Vibe for code

Code Mode takes a coding task from request to merged pull request. Sessions run in isolated cloud sandboxes, can run in parallel, persist while your machine is off, and can be triggered from third-party apps such as Slack (rolling out June 2026). The same coding agent is available through a new Vibe VS Code extension and the Vibe CLI, which adds session-scoped permissions, subagents, editable plans and a /teleport command to move a live session between terminal and cloud.

Flash Answers — speed

Le Chat’s standout everyday feature is raw speed. Flash Answers, powered by Cerebras inference hardware, delivers responses at roughly 1,000+ words per second — Mistral and Cerebras have described it as the world’s fastest AI assistant, at times an order of magnitude faster than chat experiences built on conventional GPUs.

Web search and deep research

Le Chat searches the web and cites sources, and its deep-research mode autonomously explores multiple sources to compile structured reports. Deep-research report limits are higher on Pro.

Image generation and editing

Image generation and in-conversation image editing run in partnership with Germany’s Black Forest Labs, using its FLUX models. Image editing — adjusting or restyling an existing image with a prompt — has been one of the most-praised features since its 2025 launch.

Memories

Memories (beta) carry context across conversations — preferences, decisions and references — to personalise responses. Free accounts support up to 500 memories; Pro and Team raise the limit. Memory is opt-in and editable.

Connectors and custom MCP

Le Chat ships a directory of connectors spanning data, productivity, development and commerce — Databricks, Snowflake, GitHub, Atlassian, Asana, Outlook, Box, Stripe, Zapier and more. You can also add your own via the open Model Context Protocol (MCP). Source counts vary: Mistral’s own connector page described “20+ secure connectors”, while several 2026 reviews report 40+ available on the free tier — the directory is expanding, so treat the exact number as a moving target.

Voice

Voice in Le Chat is input only, powered by Mistral’s open-source Voxtral speech model for low-latency transcription. There is no real-time conversational voice mode equivalent to ChatGPT’s.

Strengths

European data sovereignty — Le Chat hosts data in the EU by default, runs on ISO 27001-certified European data centres, and is natively subject to GDPR. For European public-sector, finance and regulated users, this is the clearest reason to choose it over US assistants (Reworked).

Speed — Flash Answers at ~1,000+ words per second makes Le Chat the fastest mainstream assistant for everyday queries.

Multilingual quality — Mistral’s models are strong across French, German, Spanish, Italian, Portuguese and other European languages, reflecting their European training focus.

Price — Pro at $14.99/month (and $5.99 for students) undercuts the $20/month standard set by ChatGPT Plus and Claude Pro.

Open weights — Mistral releases capable open-weight models (Mistral Large 3, Magistral Small, Devstral, Voxtral) under permissive licences, which the closed US labs largely do not.

Agentic breadth in one licence — the Vibe rebrand folds long-horizon work automation and remote coding into a single product and price.

Limitations

Not the frontier on hard benchmarks — Mistral’s flagship models trail Claude Opus 4.8 and GPT-5.5 on the most demanding independent coding and reasoning benchmarks. Le Chat competes on value, speed and sovereignty, not on topping leaderboards.

No video generation — unlike Gemini (Veo) or ChatGPT (Sora), Le Chat cannot generate video.

Voice is input-only — there is no real-time conversational voice mode.

Rebrand churn — the move from Le Chat to Vibe means some users will hit naming confusion, and Mistral has said it will sunset the old “Chat mode” (history is preserved in Work Mode).

Smaller ecosystem — no browser extension and a smaller third-party and plugin ecosystem than ChatGPT.

Connector counts and limits vary — figures for connectors and usage caps differ across Mistral’s own pages and third-party reviews, so verify the current limits before relying on them.

Who should use Le Chat?

Great for:

• EU and sovereignty-conscious users — data hosted in the EU by default and GDPR-native by design
• Multilingual users — particularly strong across European languages
• Speed-sensitive users — Flash Answers is the fastest mainstream assistant
• Budget and student users — Pro at $14.99 and a $5.99 student plan
• Open-source-minded developers — capable open-weight models and an MCP-based connector model
• Teams wanting agentic automation cheaply — Work Mode across business apps in one affordable licence

Look elsewhere if:

• You need the top coding or reasoning model — Claude Opus 4.8 and GPT-5.5 lead independent benchmarks
• You need video generation — use ChatGPT (Sora) or Gemini (Veo)
• You want real-time conversational voice — ChatGPT’s voice is more advanced
• You want the largest app and plugin ecosystem — ChatGPT remains broader

Historical timeline

Privacy and data handling

Mistral’s privacy posture is its central selling point, and it is materially different from the US assistants.

EU hosting by default. Mistral’s infrastructure runs on ISO 27001-certified European data centres, and data is hosted within the European Union by default. The company prioritises EU-based providers and applies GDPR-standard safeguards where any non-EU provider is involved.

GDPR-native. Mistral is subject to GDPR by design rather than by retrofit, having been built under European regulation from the start.

Training and retention.

Logging can be opt-in, and prompts can be set to auto-delete after 30 days. Le Chat Enterprise adds on-premises and private-cloud deployment for organisations with hard data-residency requirements. See Mistral’s privacy policy for full detail.

Le Chat vs alternatives

Quick comparison

Where Le Chat wins

• Data sovereignty — EU hosting and GDPR-native design
• Speed — Flash Answers is the fastest mainstream assistant
• Price — Pro and student plans undercut US rivals
• Open weights — capable models under permissive licences

Where competitors win

• ChatGPT — broader ecosystem, video generation, real-time voice, and a deeper plugin and custom-GPT marketplace
• Claude — leads independent coding and reasoning benchmarks with Opus 4.8, and offers a desktop agent (Cowork) for non-developers

For full head-to-heads, see ChatGPT vs Claude and the best AI apps ranking.

Getting started

Web

1. Go to chat.mistral.ai
2. Sign up with email, Google or Apple
3. Start in Chat or switch to Work Mode for agentic tasks

Mobile

• iOS: Download from the App Store (listed as “Vibe by Mistral (ex-Le Chat)”)
• Android: Download from Google Play

For developers

• VS Code: install the Mistral Vibe extension
• CLI: uv tool install mistral-vibe
• API: build via Mistral Studio

Tips for new users

Use Work Mode for multi-step jobs: describe an outcome — catch up on email, pull the numbers, draft the update — and approve the plan before Vibe runs it.

Connect your tools: linking Google Workspace, GitHub or SharePoint lets Vibe ground its work in your own data.

Lean on speed: Flash Answers makes Le Chat ideal for rapid back-and-forth and quick lookups.

Opt out of training if you want: enable no-telemetry mode on Pro, or use Team/Enterprise where data is excluded by default.

Frequently asked questions

Is Le Chat the same as Vibe?

Yes. Mistral rebranded Le Chat as Vibe on 28 May 2026. It is the same product with the same accounts, history, settings and plans — now organised as one agent across Work Mode and Code Mode. The mobile app is listed as “Vibe by Mistral (ex-Le Chat)”.

Is Le Chat free?

Yes. The free tier includes current Mistral models, web search, Flash Answers, image generation and editing, 40+ connectors, Memories and voice input, with EU data hosting by default. Paid Pro lifts usage limits and adds no-telemetry mode.

How much does Le Chat Pro cost?

Le Chat Pro is $14.99/month, or $5.99/month for verified students. Team is $24.99/user/month ($19.99 on annual billing), and Enterprise is custom-priced. EU customers are billed in euros at the same figures.

Is Le Chat better than ChatGPT?

It depends on what you value. Le Chat is faster, cheaper at the Pro tier, and hosts data in the EU by default. ChatGPT has a broader ecosystem, video generation, real-time voice, and stronger frontier models. For privacy and speed, Le Chat wins; for raw capability and breadth, ChatGPT leads. See ChatGPT vs Claude for the wider context.

Where is my data stored with Le Chat?

In the European Union by default. Mistral runs ISO 27001-certified European data centres and is natively GDPR-compliant. Team, Enterprise and paid API data is excluded from training by default; Pro users can enable no-telemetry mode.

Can Le Chat generate images and video?

Le Chat generates and edits images using Black Forest Labs’ FLUX models, but it cannot generate video. For video, use ChatGPT (Sora) or Gemini (Veo).

What models does Le Chat use?

Vibe runs on Mistral’s own models — Mistral Medium 3.5 for flagship work, Mistral Large 3 (open weights), the Magistral reasoning family, Devstral 2 for coding, Pixtral for vision and Voxtral for voice — plus Black Forest Labs’ FLUX for images. See the Mistral models page.

Is Le Chat good for coding?

It is capable and improving. Code Mode takes a task from request to pull request with sandboxed sessions, a VS Code extension and a CLI. On independent benchmarks, however, Mistral’s coding models trail the leaders — Claude Opus 4.8 and GPT-5.5 score higher on SWE-bench. Le Chat’s coding edge is price and integration, not benchmark wins.

Verdict

Le Chat — now Vibe — is Europe’s most credible answer to ChatGPT and Claude. It is fast, affordable, multilingual and, crucially, private by design: data sits in the EU by default and the company is GDPR-native rather than GDPR-adjacent. The 2026 rebrand to Vibe folds long-horizon work automation and remote coding into a single product, which makes it more capable than the simple chatbot it started as.

Free tier: unusually generous — current models, image generation, 40+ connectors and Flash Answers speed at no cost.

Pro ($14.99/month): the value pick among frontier assistants, undercutting the $20 standard, with a $5.99 student rate that is hard to beat.

Team and Enterprise: the strongest choice for European organisations that need data residency, no-training guarantees and on-premises options.

Le Chat will not top independent coding or reasoning benchmarks — Claude Opus 4.8 and GPT-5.5 lead there, and it lacks video and real-time voice. But for speed, price, multilingual quality and European data sovereignty, it is the clear pick, which is why it topped France’s App Store on launch and why Mistral is now reportedly valued near €20 billion.

Official links