Mistral Le Chat adds Deep Research Mode, voice input, Think Mode, Projects and more

Mistral AI has released a major update to its AI Chatbot Le Chat, with Deep Research Mode as its standout addition. This new mode allows the chatbot to break down complex questions, search across multiple sources, and generate structured reports with references. While comparable to similar tools in ChatGPT, Google Gemini, Claude, or Grok, Mistral cautions that outputs may contain inconsistencies, making it more useful for quick overviews than for detailed fact-checking.

Le Chat also now supports real-time voice input powered by the recently launched Voxtral model, enabling natural conversations and voice-based note-taking. For tasks requiring deeper logic, the new Think Mode uses the Magistral model to handle step-by-step reasoning.

The update also introduces a revamped image editing tool developed with Black Forest Labs, allowing users to modify generated visuals using text commands, likely using Kontext models to preserve coherence. Finally, a new Projects workspace lets users group chats, files, and tools into persistent workspaces, with autosave and file upload support. Each project retains its tools, settings, and context, making it easier to manage ongoing work without losing track of configurations or ideas across sessions. All features are now available on the web, iOS, and Android versions of Le Chat.