Mistral

Mistral AI is a French AI company that builds and publishes both open-weight and proprietary large language models, along with an enterprise AI platform for customizing, fine-tuning, and deploying AI assistants, agents, and multimodal models. It was founded in April 2023 in Paris by Arthur Mensch (CEO, a former Google DeepMind researcher), Guillaume Lample, and Timothée Lacroix (both former Meta AI researchers). The company raised a €1.7 billion Series C in September 2025 at an €11.7 billion post-money valuation, led by ASML (which became Mistral's largest shareholder at roughly an 11% stake), and secured $830 million in debt financing in March 2026 to fund a new Nvidia-powered data center near Paris; as of mid-2026 the company was reportedly in early talks to raise roughly €3 billion more at a valuation near €20 billion, though that round hadn't been confirmed closed.

Mistral's current flagship model is Large 3, a 675-billion-parameter mixture-of-experts model (about 41 billion active parameters per token) released as open weight under the Apache 2.0 license. On May 28, 2026, Mistral rebranded its consumer chatbot, Le Chat, as Vibe, repositioning it from a chatbot into a unified AI agent platform with two modes: Work Mode (connecting to Google Workspace, Outlook, SharePoint, and Slack to execute multi-step tasks autonomously after user approval) and Code Mode (connecting to GitHub and carrying a coding task through to a pull request). Vibe also includes web search, a canvas editor, image generation via Black Forest Labs' Flux Ultra, a code interpreter, document OCR, and a no-code agent builder, and is available on web, iOS, and Android.

1. Sign up at chat.mistral.ai (or download the iOS/Android app) to use Vibe (formerly Le Chat) — the Free plan requires no payment.
2. Chat directly for research, writing, and general Q&A, or switch on Work Mode to connect Vibe to Google Workspace, Outlook, SharePoint, or Slack and have it carry out a multi-step task after you approve the plan.
3. For coding tasks, use Code Mode: connect a GitHub repository and describe what you want built or fixed — Vibe carries the work through to a drafted pull request.
4. Use built-in tools as needed: web search for up-to-date information, the canvas editor for iterative document/content editing, image generation (via Flux Ultra), the code interpreter for data/analysis tasks, and document OCR for extracting text from files.
5. For custom or no-code agents, use Vibe's agent builder to define a specific, repeatable assistant without writing code.
6. For programmatic or enterprise use, access Mistral's models directly through the Mistral AI platform/API — including the open-weight Large 3 — or deploy on your own infrastructure given Mistral's open-weight licensing, and upgrade from Free to Pro, Team, or Enterprise as usage and collaboration needs grow.

• General chat, research, and writing assistance through Vibe (formerly Le Chat), on web, iOS, or Android.
• Automating multi-step work tasks — drafting an email thread, updating a shared doc, coordinating a Slack update — via Work Mode's connections to Google Workspace, Outlook, SharePoint, and Slack.
• Coding tasks carried from a description through to a drafted pull request via Code Mode's GitHub connection.
• Enterprise deployment of AI assistants and agents built on Mistral's models, for organizations that need customization, fine-tuning, or on-premises/self-hosted deployment using the open-weight models.
• Building a no-code custom agent for a specific, repeatable task without writing code.
• Multimodal work — document OCR, image generation, and a code interpreter for data analysis — inside the same chat interface.