The Vibe Interface

Updated November 2, 2026 — Le Chat is now Vibe. As of May 28, 2026, Le Chat is now Vibe (still at chat.mistral.ai): a single agent for work and code, carrying over your conversations, settings, and subscriptions. The legacy Le Chat experience remains accessible in Chat mode; in Work mode, agents are now skills (invoked with @), memories are the knowledge base, and deep search is the Deep Research skill. Plans are now called Mistral Free, Pro, Team, and Enterprise. This lesson uses the name Vibe; some menus described here may have changed names or locations.

Mastering the Vibe Interface

You now understand the models and core concepts; it’s time to become an effective Vibe user. The difference between someone who has “tried AI” and someone who actually uses it often comes down to a deep knowledge of the interface: knowing where to find an answer from three days ago, when to switch models mid-stream, and how to avoid repeating the same instructions in every conversation. Let’s walk through each element.

The sidebar, on the left side of the screen, is your navigation hub, and its core is the conversation history. All your past conversations are listed there, sorted by date: you can resume a conversation by clicking on it, rename it for easier retrieval, delete it when it’s no longer useful, or search across all your exchanges. This history is more valuable than it seems. If you got an excellent rewrite of a difficult paragraph last week, you can find it in seconds instead of asking again—and risking a worse answer.

At the top of the sidebar is the new conversation button. Each new conversation starts with a blank slate: the model doesn’t remember your previous exchanges, unless you use custom instructions, which we’ll cover below. This is intentional, not a limitation—it prevents an old topic from contaminating the new one. Finally, depending on your plan, you can organize your conversations into thematic spaces: one for work, one for personal projects, one for research. Anyone juggling report preparation, technical monitoring, and ad hoc questions benefits greatly from this separation.

The Model Selector

At the top of the conversation area or near the input field, the model selector lets you choose which model will handle your request. The habit to build is simple: Mistral Small 4 (or the default option) covers most daily exchanges; Mistral Large 3 is necessary when you need the highest possible quality, for example for a document that will be read by your leadership; and for reasoning and analysis problems—where logic matters more than style—Small 4’s reasoning mode activates automatically.

Often overlooked detail: you can switch models mid-conversation, and the new model will have access to the entire exchange history. Nothing stops you from exploring a topic with Small 4, then switching to Large 3 for the final version, without re-explaining anything.

The Input Area

This is the main field where you type your messages, and it does more than just accept text. You can attach files directly: images, which multimodal models like Small 4 and Large 3 know how to interpret (a photo of a whiteboard after a meeting, for example, that you ask to transcribe); PDF documents to analyze or summarize; text files for data or code processing; spreadsheets for data analysis.

Three keyboard shortcuts are worth memorizing:

Enter: send the message
Shift + Enter: go to a new line without sending
Up Arrow: recall the last sent message to edit it

The second one is a game-changer for new users: it lets you structure a prompt across multiple paragraphs without sending it prematurely.

Settings

The settings icon—usually a gear—opens customization options, and by far the most powerful of these is custom instructions. Here you define guidelines that apply to all your conversations: “Always respond in French,” “Use a professional tone,” “Give concise answers,” or “I am a Python developer, adapt your code examples.” These instructions work like a permanent system prompt that steers the model’s behavior in every exchange. Concretely, if you catch yourself repeating the same clarification at the start of every conversation, that’s a signal it should move to custom instructions.

Settings also group display preferences—dark or light mode, font size, conversation area width—and account management: profile information, current plan and billing, logout.

Advanced Features

Three features deserve an introduction, even though the last lesson of the course will cover them in more detail.

Deep Research, available with the Pro plan, lets the model browse the web thoroughly. Unlike a simple web search, Deep Research consults multiple sources, cross-references information, synthesizes results, and provides cited sources. Enable it when you need up-to-date factual information or exhaustive research—an overview of a topic you’re discovering, for example—and keep simple search for quick verifications.

Canvas is a collaborative workspace integrated into Vibe. It lets you write and edit text in collaboration with the model, write and debug code in an integrated editor, and iterate on a document without polluting the conversation history. Whenever a text requires more than two or three back-and-forths—an article, a report, a script—Canvas becomes more comfortable than the classic exchange.

Finally, Agents are pre-configured versions of Vibe, optimized for specific tasks: writing professional emails, data analysis, translation, code development. Mistral and the community offer them, and you can create your own with custom instructions—a “review my newsletters” agent configured once saves you from restating your style guidelines for every issue.

Optimizing Your Daily Use

A few habits make a difference over time. Dedicate one conversation per topic or project, rename them with explicit titles, and delete test conversations: a clean history is consultable, a messy history is a burden. Use context wisely: in an existing conversation, the model remembers the entire exchange, which is valuable as long as you stay on the same topic; as soon as you change topics, create a new conversation to avoid interference. Finally, combine models: start with Small 4 for initial exploration, and switch to Large 3 if the answer lacks depth—for pure logic, Small 4’s reasoning mode does the job without changing models.

Key Takeaways

The sidebar organizes your conversations and lets you find them quickly
The model selector lets you adapt power to your need, even mid-conversation
Custom instructions steer the model’s behavior for all your conversations
Deep Research and Canvas are advanced features for research and collaboration
Organize your conversations by topic and name them clearly to gain productivity