Arthur Mensch

Mistral OCR 3 extracts text and images from documents with SOTA accuracy. It handles handwriting, complex tables, and scanned forms, outputting clean markdown.

Devstral 2 is the new SOTA open-weight coding family, achieving 72.2% on SWE-bench Verified. It ships with Mistral Vibe, an open-source CLI agent for end-to-end code automation. Currently free via API.

Mistral 3 includes three state-of-the-art small, dense models (14B, 8B, and 3B) and Mistral Large 3 – our most capable model to date – a sparse mixture-of-experts trained with 41B active and 675B total parameters. All models are released under the Apache 2.0 license. The Ministral models represent the best performance-to-cost ratio in their category. At the same time, Mistral Large 3 joins the ranks of frontier instruction-fine-tuned open-source models.

Create AI use cases, manage the full lifecycle, and ship with confidence, all with enterprise privacy, security, and full ownership of your data.

Mistral's Le Chat gets a major upgrade with Deep Research, a new Voice Mode, Projects for organizing chats, and advanced image editing. This update bundles powerful new capabilities into Mistral's core AI chat assistant.

Voxtral by Mistral AI is a new family of open-source speech understanding models. Available in 24B and 3B sizes, it goes beyond transcription to offer Q&A, summarization, and function calling directly from voice with SOTA performance.

Magistral is Mistral AI's first reasoning model, excelling in domain-specific, transparent, and multilingual reasoning. Available as an open-source 24B version (Small) and a powerful enterprise version (Medium).

Mistral Code is an enterprise AI coding assistant with powerful models (Codestral, Devstral) & IDE integration. Offers customization, on-prem deployment & full stack control.

Mistral AI's new Agents API helps developers build AI agents with persistent memory, tool use (code execution, web search, image gen, MCP) & orchestration. Simplifies complex, actionable AI workflows.

Secure, customizable, agent-powered AI assistant for business. Integrates company knowledge (Drive, etc.), build custom agents. Deploy cloud or on-prem.

Introducing Mistral Small 3.1. Multimodal, Apache 2.0, outperforms Gemma 3 and GPT 4o-mini.

Introducing Mistral OCR – a cutting-edge, lightweight Optical Character Recognition model designed for speed, accuracy, and efficiency. Whether extracting text from images or digitizing documents, it delivers state-of-the-art performance with ease.

Mistral Saba is a regionally-focused AI model (24B) from Mistral AI, excelling in Arabic and many Indian languages. Outperforms larger models in its target region. API + local deployment.

Mistral Small 3 is the most efficient and versatile model of Mistral. Pre-trained and instructed version, Apache 2.0, 24B, 81% MMLU, 150 token/s. No synthetic data so great base for anything reasoning.

Le Chat combines powerful AI with extensive information on the web to help you rediscover the world. Enjoy natural conversations, real-time internet search, comprehensive document analysis, and much more.

Compared to its predecessor, Mistral Large 2 is significantly more capable in code generation, mathematics, and reasoning. It also provides a much stronger multilingual support, and advanced function calling capabilities.

Choose between Mistral Small, Mistral Large and Mistral Next. Mistral Large supports English, French, Spanish, German and Italian.

Mistral Large has top-tier reasoning capacities, is multi-lingual by design, has native function calling capacities and a 32k model. The pre-trained model has 81.2% accuracy on MMLU.

Mistral 7B is a 7.3B parameter model that:

• Outperforms Llama 2 13B on all benchmarks
• Outperforms Llama 1 34B on many benchmarks
• Approaches CodeLlama 7B performance on code, while remaining good at English tasks