Voice Mode

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 now uses the name Vibe; some menus described here may have changed names or locations.

Speak naturally to your assistant

Voice Mode turns your written exchange with Vibe into a continuous oral conversation. It relies on Voxtral, Mistral AI’s speech recognition model: your speech converts to text, gets processed, and the response reads back to you aloud. The benefit isn’t just comfort—it opens up usage moments you’d otherwise lose, like driving, cooking, or walking between meetings.

Activation requires no setup. On a smartphone, it takes three steps:

1. Open the Vibe app on your phone
2. Tap the microphone icon in the input bar
3. Speak naturally—Vibe listens and responds vocally

On a computer, the principle is identical: the same microphone icon in the web interface starts a voice session.

What Voxtral does

Voxtral is Mistral AI’s speech-to-text model. It transcribes your speech with good accuracy, even when the environment isn’t silent, and works in real time: no perceptible latency between the end of your sentence and the start of the response, which is essential for a conversation to feel like a conversation.

Language coverage is broad—French, English, Spanish, German, Italian, Portuguese, Dutch, Russian, and Chinese—and crucially, it doesn’t lock you in. You can switch languages mid-conversation: Voxtral detects the spoken language and adapts without you having to toggle a setting. On its end, the response comes through a natural synthetic voice, and multiple voices are available in settings if the default one doesn’t suit you.

Situations where voice makes a difference

Brainstorming on the move is the most revealing use case. An idea strikes while you’re walking: instead of jotting it down halfway and forgetting it, you say it out loud and Vibe helps you structure it, spot weak points, and suggest additional avenues.

Interview prep works on the same principle. Ask Vibe to play the role of the recruiter, answer orally as you would on the actual day, then listen to its evaluation and advice: you work on substance and phrasing in real-world conditions, which a written exchange doesn’t allow. Language learning benefits from the same setup—converse in the target language, ask for real-time corrections, and adjust the level as you progress.

More practically, dictation remains a significant time saver. Meeting notes, a report, ideas tossed down after a meeting: dictate them, Vibe transcribes them, and can reorganize them into a structured document right away. Finally, voice has an accessibility aspect you shouldn’t underestimate: it makes Vibe usable for people struggling with a keyboard, whether due to a disability, temporary mobility issues, or hands full.

Speak clearly to be understood

Voxtral handles ambient noise, but you’ll get better transcriptions by articulating at a natural pace, without mumbling or rushing. Prompt quality matters just as much as in text: saying “First, summarize the article I just shared. Then, identify the three main arguments. Finally, propose a constructive critique” yields an organized response where “what do you think?” yields vagueness.

Nothing forces you to stay in voice from start to finish. Hybrid mode is often the most productive: dictate the initial request, then switch back to text to refine a detail, paste a link, or correct phrasing. Finally, make it a habit to proofread the transcription after a long dictation—proper nouns, technical terms, and acronyms are the weak points of any speech recognition, and a few seconds of correction saves you from building an entire document on a misheard word.

Try it now, before reading further: activate Voice Mode on your phone and ask an open question orally, such as “What are the pros and cons of remote work for a small business?” Listen to the response, follow up on a specific point to check that the thread holds, then follow up with a question in English—the language switch happens without any action.

Where voice shows its limits

Returning to the keyboard is necessary in several cases. Highly technical queries first: dictating code, a math formula, or a data table is laborious and error-prone. A very noisy environment then degrades recognition accuracy, despite the model’s robustness. Long, context-heavy conversations are better managed in text, where you can review and reference what was said. Then there’s confidentiality, often decisive: in an open office or on public transit, speaking aloud about professional topics isn’t an option.

Key takeaways

• Voice Mode uses Voxtral for real-time speech recognition
• Multiple languages are supported with automatic detection
• Ideal for brainstorming on the move, dictation, and accessibility
• You can alternate between voice and text in the same conversation
• Text remains preferable for highly technical tasks or noisy environments