Mistral AI Unveils New Chat Features and AI Models

French AI startup Mistral has announced updates to its Le Chat platform and introduced a new multimodal AI model called Pixtral Large.

These updates aim to enhance user experience and keep the company competitive in the AI space.

Le Chat now includes several new features:

• Web Search with Citations: Users can search the web directly within the chat, with sources cited in responses.
• Canvas Tool: Inspired by ChatGPT Canvas, this feature allows users to create and edit content like presentations, documents, and webpage mockups. Users can modify content in place, version drafts, and preview designs.
• Document and Image Processing: The chatbot can analyze and summarize large PDFs and images, including charts and equations.
• AI Agents: Shareable workflows automate tasks like scanning expense reports and processing invoices.

Introducing Pixtral Large

Mistral’s new model, Pixtral Large, is a multimodal AI capable of understanding text and images. It is part of Mistral’s Pixtral model family and features 124 billion parameters, making it one of the most powerful models in the market. It competes with leading models like Anthropic’s Claude 3.5 Sonnet, Google’s Gemini 1.5 Pro, and OpenAI’s GPT-4o on benchmarks.

Pixtral Large excels in processing documents, charts, and natural images, demonstrating advanced capabilities in image understanding.

Free Beta Features

All new features on Le Chat, including web search, canvas, document processing, and AI agents, will remain free during the beta phase.

With these updates, Mistral continues to push the boundaries of AI technology, providing users with advanced tools for a variety of tasks, from content creation to data analysis.