AI chat and coding agent from Mistral, formerly Le Chat
Mistral Vibe, renamed from Le Chat in May 2026, is Mistral's conversational AI and coding assistant for consumers, developers and professionals.
AI Panel Score
6 AI reviews
Reviewed
AI Editor ApprovedApproved and published by our AI Editor-in-Chief after full panel analysis.Mistral Chat is a conversational AI interface for interacting with Mistral's family of large language models through the browser, supporting multi-turn conversations and switching between model versions. Pricing is freemium: a free tier includes image generation and more than 40 connectors, the Pro plan costs $14.99 per month, and Team seats run $24.99 per month. Key capabilities include web search with citations, a Deep Research mode, a code interpreter, a canvas editor, voice mode, chain-of-thought Think mode, persistent memories, and custom agents with shareable knowledge bases. Work Mode connects the assistant to email, calendar, and documents for workflow automation. TopReviewed's six-seat AI review panel scored it 7.7/10, praising the availability of Deep Research and the code interpreter at the $14.99 Pro tier while noting brand recognition lags ChatGPT and Claude in enterprise buying conversations. It best fits teams that want comparable capability at lower per-seat cost with European data handling.
Users access Mistral Vibe through a web browser, type prompts or questions into a chat window, and receive generated responses from Mistral's language models. Conversations are maintained in a session history visible in the sidebar, allowing users to return to previous threads. The interface is straightforward: no setup or local installation is required to begin a conversation.
The platform runs on Mistral's current models, including Mistral Medium 3.5, Mistral Small 4, Mistral OCR 4 and Voxtral for voice, and supports capabilities such as code generation, document summarization, and general question answering. A file upload feature allows users to attach documents for the model to reference when generating responses. The interface also includes a canvas-style mode for working on longer-form written content iteratively.
Mistral Vibe is aimed at general consumers, developers exploring Mistral's models, and professionals looking for a hosted AI assistant without API integration overhead. A free tier is available with access to a baseline model. A paid Pro subscription is priced at $14.99 per month and unlocks access to more capable models, higher usage limits, and additional features. Comparable products in the category include ChatGPT by OpenAI, Claude by Anthropic, and Gemini by Google.
Mistral Vibe runs in the browser with no install or API key, and native iOS and Android apps give the same chat, voice and image-generation features on mobile. It connects to over 100 tools including email, calendar, Slack, GitHub and Jira through connectors and full MCP compatibility. For developers who want programmatic access rather than a chat interface, Mistral offers a separate API product. The chat interface itself does not require API key configuration from the user.
A sandboxed code execution environment that handles data documents (Excel, CSV, JSON), answers quantitative questions, corrects data errors, and generates data visualizations by running Python code.
Autonomously plans a research strategy, navigates the web to synthesize multiple trusted sources, and generates long-form structured analytical reports on complex topics.
Generates photorealistic or creative images from text prompts using Black Forest Labs' Flux Ultra model, and supports iterative prompt-based editing that preserves characters and design elements across edits.
Provides step-by-step reasoning powered by Mistral's thinking models for problems that require deeper reflection, enabling chain-of-thought responses to complex queries.
Allows users to build reusable AI assistants with custom system instructions, attached tools (including MCP connectors), and linked knowledge bases that can be shared securely with a team.
An agentic mode for multi-step tasks that connects to mailboxes, calendars, documents, and other enterprise tools simultaneously, enabling users to catch up on emails, run structured web research, and produce editable briefs — available on Pro, Team, and Enterprise plans.
An integrated side-by-side writing and coding surface where users can collaboratively refine AI-generated text, code, or presentations in real time, including versioning drafts and previewing designs.
Lets users group related chats into focused project spaces with their own default document libraries, persistent tool settings, and uploaded files — keeping conversations, documents, and ideas organized in one place.
Supports multi-turn conversations with Mistral's latest models, retaining context across exchanges with multilingual support and automatic language detection.
Allows users to speak instead of type using Mistral's Voxtral audio model, which handles transcription and supports multiple languages.
Searches the web in real time and combines pre-trained model knowledge with recent information from journalism, social media, and other sources to deliver cited, evidence-based responses.
An opt-in feature that lets Mistral Vibe learn and remember user preferences across sessions, enabling personalized responses, goal tracking, and continuity over long-running workstreams.
Native Mistral Vibe applications available on the App Store and Google Play Store, offering access to all core features including chat, voice mode, and image generation on mobile devices.
Supports self-hosted, private cloud, or Mistral-hosted deployment with SAML SSO integration, strict access control list (ACL) adherence for data connections, and full administrative control over team roles, billing, and security policies.
Personal AI assistant for everyday life and work.
Enhanced productivity, higher limits, and Mistral Vibe for all-day development.
Secure, collaborative workspace for teams building with AI.
Private deployments powered by custom models, UI, and tools.
Mistral punches above its weight at $14.99, but ChatGPT still owns the room.
“Founded April 2023, Mistral AI has shipped a genuinely competitive product with Deep Research, Code Interpreter, and Work Mode in three years. At $14.99/month Pro, the value math is hard to argue with.”
Series C European AI lab, shipping fast. The changelog shows native iOS/Android apps, Voxtral voice mode, MCP connectors, and image generation via Black Forest Labs — all live. That's not a roadmap. That's execution. For a company three years old, the surface area is impressive.
The tradeoff worth naming: Mistral isn't ChatGPT in mindshare or ecosystem. OpenAI has the enterprise relationships, the brand recognition, and the board-level familiarity. If your team is already on ChatGPT Teams, switching costs aren't zero. Mistral's differentiation is European data residency, open-weight model heritage, and 6x message limits over free at $14.99 — real advantages for specific buyers.
Two things I'd watch. One: the €1.7B Series C at an €11.7B valuation buys runway — and growth expectations that reach pricing. Two: Work Mode and Custom Agents are Pro/Team-only features — the capabilities that matter most are paywalled. Pilot Pro for 90 days before any org-wide commitment.
Trails ChatGPT and Claude in enterprise mindshare, but $14.99 Pro vs comparable tiers and European data options create a real wedge.
European AI credibility is a legitimate board talking point, especially for compliance-sensitive organizations.
Browser-based, no install, free tier live immediately — Pro at $14.99 unlocks Deep Research and 15GB storage with zero IT lift.
Work Mode's multi-step agentic connections to email, calendar, and docs advances productivity beyond simple Q&A.
Founded April 2023, €1.7B Series C at an €11.7B valuation, shipping weekly — funded well past the near-term horizon.
Teams that want ChatGPT-level capability at $14.99 and care about European data handling.
Your org is already standardized on OpenAI and switching costs outweigh the pricing delta.
Mistral Chat is a serious ChatGPT challenger with real enterprise scaffolding at $14.99.
“Mistral Vibe has grown well past a simple chat wrapper — Custom Agents, Work Mode, and Enterprise SSO signal a platform with genuine team adoption potential. The gap vs. ChatGPT and Claude isn't feature count; it's brand trust and entrenched user habits.”
The feature surface here is broader than most CS teams expect from Mistral. Work Mode's agentic connections to mailboxes and calendars, the 40+ enterprise connectors on the free tier, and Projects with persistent document libraries — that's a workflow architecture, not a chatbot. For a CS team running onboarding playbooks, escalation briefs, or knowledge base drafts, this toolset covers the daily loop. Persistent Memories across sessions is particularly relevant for teams where continuity of context is a retention asset.
The Team plan at $24.99/seat with 30GB storage, data export, and Admin API is where I'd evaluate this against Slack-adjacent tooling and Claude for Teams. The tradeoff is adoption friction — Mistral's brand recognition in CS org buying conversations still trails OpenAI and Anthropic, which means you're selling the tool twice: once to procurement, once to your reps.
If we adopt this at the team tier today, in three years we either have a deeply embedded workflow tool with custom agents trained on our CS knowledge base, or we've re-onboarded to Claude because the category consolidated. The SSO, audit logs, and white-label Enterprise options suggest Mistral is building for durability — but I couldn't find a public roadmap, and that's the gap I'd want answered before committing org-wide.
Priced below Claude for Teams and competitive with ChatGPT Team, but Mistral's European AI identity is a differentiator for data-residency-sensitive enterprise buyers, not a mainstream CS buying trigger yet.
Work Mode's email, calendar, and document integrations map directly to CS workflows, though I couldn't find a native CRM connector to close the loop on accounts and tickets.
40+ enterprise connectors on the free tier and MCP support on Custom Agents give solid surface area, though the changelog tracks models more closely than integrations.
Enterprise SSO, audit logs, and Admin API on the Enterprise tier show infrastructure built for organizational lock-in — in the right way — but Mistral's market position in 36 months is a real open variable.
Deep Research Mode, Think Mode, Custom Agents with MCP connectors, and Canvas versioning indicate a platform team thinking beyond chat primitives.
CS teams at mid-market companies that want ChatGPT-tier capability with stronger data controls and a lower per-seat cost.
Your CS org is already standardized on OpenAI or Anthropic and your CRM integration is non-negotiable before tool adoption.
$14.99/month Pro tier with 6x message limits — honest pricing in a murky category
“Four tiers, all publicly priced except Enterprise. SSO lives at Enterprise — that's the usual tax.”
Free tier includes image generation and 40+ enterprise connectors — unusually generous at $0. Pro at $14.99/month unlocks 6x message volume and 30x deep research capacity. Team at $24.99/month adds 30GB storage and data export. Compare to ChatGPT Plus at $20/month — Mistral undercuts by $5 with comparable feature breadth based on their pricing page.
50-seat Team deployment: $24.99 × 50 × 12 = $14,994/year. Add 20% seat creep by year 2. Year 3 lands around $21,600. Vibe coding past plan limits bills PAYG at published token rates — session burn is the unmodelable part. SSO requires Enterprise negotiation; no published Enterprise price.
Monthly billing visible, no minimum term indicated on pricing page. Contract flexibility looks standard SaaS. ROI is measurable on productivity tasks — Deep Research Mode and Code Interpreter have concrete time-displacement logic — but no published benchmark data to anchor it.
Team tier includes Admin API and data export, reducing procurement friction; SSO gated to Enterprise adds negotiation overhead for security-conscious buyers.
Monthly billing implied by pricing page; no minimum term or auto-renewal window published, which is either clean or undisclosed.
Three of four tiers fully published with feature detail; Enterprise pricing requires a sales conversation, which is category norm.
Deep Research Mode and Code Interpreter map to measurable time savings, but no published benchmark or case study data to validate displacement claims.
Vibe coding reverts to PAYG beyond Pro limits at published token rates, but session-level consumption is the year-3 modeling gap for developer-heavy teams.
Teams of 10-50 needing a full-featured AI assistant at $24.99/seat without SSO requirements.
Your security policy mandates SAML SSO — Enterprise negotiation adds cost and time you can't model upfront.
Work Mode is the daily win — and it sits behind the Pro wall
“Mistral Chat has quietly built a surprisingly complete stack: deep research, code interpreter, canvas editor, voice mode, image gen, and Custom Agents all in one browser tab. The free tier is generous enough to evaluate seriously before paying.”
Forty-plus enterprise connectors on the free plan is a genuine differentiator. Most support agents I know are copy-pasting between tools all day — having mailbox and calendar connections in a chat interface matters. Work Mode looks like the killer feature for that workflow, but it's locked to Pro ($14.99/mo) and up. That limit will sting anyone who evaluates on free and then hits the wall.
The Projects and Libraries structure is the right mental model for support work — ticket categories, product lines, incident threads. Persistent Memories across sessions is opt-in, which is the correct call for data-conscious teams. Deep Research Mode with citations addresses the 'where did you get that?' problem that makes ChatGPT answers hard to trust in escalation docs.
The friction I'd watch: the changelog reads like model release notes, so what changed in the chat app between last Tuesday and today is guesswork. Canvas and code interpreter are powerful but buried — not obvious in a sidebar-heavy interface. Claude and ChatGPT both surface advanced features more aggressively. Day three, non-power-users will still be typing into the main box and missing half the product.
Projects, memories, and web search citations reduce daily copy-paste friction, but Work Mode's Pro paywall means free-tier users hit a feature gap fast.
Docs, API reference, and a dated changelog are all public, though the changelog is written for developers rather than daily chat users.
New features land without in-product signposting; canvas and code interpreter aren't front-and-center, so they quietly go unused by most daily users.
Custom Agents with MCP connectors, Think Mode chain-of-thought, and up to 1,000 projects on Pro signal real power-user investment beyond surface-level chat.
40+ enterprise connectors and calendar/mailbox access via Work Mode suggest real workflow design thinking, not just a chat box with a nice skin.
Support agents or small teams who want a single browser tab replacing three tools and don't mind paying $14.99 to unlock the agentic features.
Your team needs SAML SSO and audit logs, because those only arrive with an Enterprise negotiation.
A deceptively simple surface hiding a week of features
“Mistral Vibe Pro has quietly grown into a genuinely competitive AI assistant. Most people will miss half the features because the surface looks deceptively simple.”
First impression is clean and low-friction. No install, no API keys, no setup ritual. You land in a chat window and go. The free tier even includes image generation and up to 500 persistent memories, which is more than ChatGPT hands you at zero dollars. That's a strong opening move.
Day three is where it gets interesting. Deep Research Mode, Code Interpreter, Canvas, Custom Agents, Work Mode — that's a real feature stack. The problem is discoverability. Nothing in the interface obviously says 'hey, you can build reusable agents with MCP connectors.' You have to find it. Month three users who dig will feel rewarded. Month-one users will probably underuse it and wonder why they're paying $14.99.
The native iOS and Android apps fix what would've been a real gap. 'Mobile parity' isn't just a checkbox here — Voice Mode via Voxtral works on mobile too. The tradeoff is that Mistral's brand recognition is still well below Claude or Gemini, so the onboarding assumes more curiosity than average users bring.
Canvas editor and Projects feel considered; the changelog is developer-facing, so quiet improvements to the app are easy to miss.
Day one is easy; day thirty requires deliberate exploration to find Work Mode, Deep Research, and 1,000-project limits that Pro actually unlocks.
Dedicated iOS and Android apps with Voice Mode via Voxtral and image generation close what could've been a wide gap against ChatGPT's mobile experience.
Zero-setup browser access is great, but a feature set this deep — Custom Agents, Work Mode, MCP connectors — needs better surface-level signposting for new users.
Browser-only historically feels fragile, but native iOS and Android apps and 15GB document storage on Pro suggest infrastructure the team is actually maintaining.
Professionals and developers who want a capable AI assistant at $14.99 and are willing to spend a week actually learning what it can do.
You want a tool that shows you everything upfront and never makes you go hunting.
Solid European challenger, but ChatGPT's shadow is long
“Mistral Chat is a genuinely capable freemium chatbot with a surprisingly deep feature set for $14.99/month. The differentiation story is thin against OpenAI and Anthropic, but the European provenance and open-model heritage give it real footing.”
Three tells. One: the changelog is model-and-API shaped — the chat product's changes surface on the news blog instead, which is an odd split. Two: founded April 2023, which means three years of product history. Three: the free tier includes 40+ enterprise connectors, which is either genuinely generous or a number that ages poorly if limits tighten.
Feature breadth is real. Code Interpreter, image generation via Black Forest Labs' Flux Ultra, Think Mode, custom agents with MCP connectors — this isn't a stripped-down ChatGPT clone. Work Mode agentic integration is a legitimate differentiator against Claude's current offering. The $24.99 Team tier with data export and Admin API signals they're chasing enterprise, not just consumers.
Exit portability is decent. Conversations are standard text. No proprietary format lock-in visible. The tradeoff: if you build deeply on custom agents or persistent memories, migration gets messier. Category norm is that memory portability is always worse than it looks.
Work Mode and European open-model lineage are real edges, but ChatGPT and Gemini match or exceed most core features; differentiation is narrow.
Browser-based text chat is inherently portable, but custom agents and persistent memories create soft lock-in that won't be clean to migrate.
Mistral AI has named VC backing and EU regulatory tailwinds; Enterprise tier with SAML SSO and custom deployments signals a serious business model, not a hobby project.
Feature descriptions are specific and grounded — Flux Ultra is named, limits like '6x Free' are quantified — no egregious superlatives visible.
Three years in with a dated public changelog and steady feature depth; matches the early-stage survivor pattern, not the failure pattern — so far.
Professionals and teams who want a capable ChatGPT alternative with European data residency options and don't need OpenAI's ecosystem integrations.
You're already deep in the ChatGPT or Claude ecosystem and need a compelling reason to switch — there isn't one here yet.
Common questions answered by our AI research team
Yes. The web client at chat.mistral.ai needs no install, and native iOS and Android apps cover the same chat, voice and image features on mobile.
Yes, Mistral Chat supports multi-turn conversations, meaning it retains context across multiple messages within a session.
Users can switch between different Mistral model versions within the platform.
Who Mistral Vibe competes with, and how each one differs
Closed-weight model provider with GPT-4/GPT-4o series and ChatGPT consumer app; larger enterprise adoption and no open-weight release strategy, unlike Mistral's open-weight models.
Closed-weight API-only provider (Claude models) focused on safety-oriented enterprise deployments, with no open-weight model releases as Mistral offers.
Open-weight model family backed by Meta's infrastructure and free licensing for most commercial use, competing directly with Mistral's open-weight releases but without a first-party commercial API platform.
Closed-weight, multimodal-first models tightly integrated with Google Cloud and Workspace, targeting large enterprises rather than the open-weight developer community Mistral emphasizes.
Hosting and fine-tuning platform for open-weight models (including Mistral's own) rather than a model developer, competing on inference infrastructure and pricing rather than proprietary model quality.
Model hub and MLOps ecosystem hosting thousands of open-weight models from many labs, competing with Mistral's distribution channel but not producing a single flagship model family or enterprise API at the same scale.
Company
Mistral AIFounded
2023Pricing
From $14.99/moFree Plan
AvailableMistral AI is a Paris-based AI company that develops and distributes open-weight large language models and offers a commercial API platform for enterprise deployments.
