Ce que Mistral AI a publié en avril 2026

Remote agents in Vibe.Powered by Mistral Medium 3.5.

Mistral introduced Mistral Medium 3.5, a 128B-parameter model with a 256k context window, now the default in Le Chat and Vibe. It enables remote coding agents that run in the cloud, handling long-hori…

Mistral Medium 3.5

Mistral AI launched Mistral Medium 3.5, a frontier-class multimodal model optimized for agentic and coding use cases, released under a Modified MIT license. The model is positioned as a significant up…

Mistral AI released Mistral Medium 3.5, a frontier-class multimodal model optimized for agentic and coding use cases, under a Modified MIT license. The model is positioned as a frontier-class solution

Workflows for work that runs the business

Mistral AI introduced Workflows in public preview as an orchestration layer for enterprise AI, addressing reliability gaps in production AI pipelines. The tool enables durable execution, observability…

AXA revolutionizes the future of insurance with AI.

AXA is integrating Mistral AI's models into its Secure GPT platform to empower employees and develop tailored AI use cases for insurance operations. The collaboration aims to reduce bias, improve inte…

Ardian leverages Mistral AI to support due diligence and deal analysis workflows.

Ardian, a $192B private investment firm, partnered with Artefact and Mistral AI to develop GAIA, an internal generative AI platform for due diligence and deal analysis. The platform processes unstruct…

ASML and Mistral.

Mistral announced a partnership with ASML to integrate frontier AI across ASML’s lithography systems, R&D, and operations. Initial use cases include AI agents for software reliability, log analysis fo…

HSBC and Mistral.

HSBC announced a partnership with Mistral AI to boost productivity and enhance AI capabilities across the bank. The collaboration aims to equip HSBC employees with AI tools to innovate, simplify tasks…

Advancing defence AI: How DSO trained a custom MoE model.

Mistral partnered with DSO National Laboratories to co-develop a custom Mixture-of-Experts (MoE) large language model optimized for Southeast Asian languages, outperforming Llama 3 on regional benchma…

BigDataCorp unlocks advanced AI-driven insights with Mistral on AWS Bedrock.

BigDataCorp, a data analytics firm, partnered with AWS Bedrock to integrate Mistral models, enhancing its AI-driven insights for clients. The integration enabled advanced NLP and ML capabilities, allo…

Belfius enhances customer experience with a multi-lingual chatbot.

Belfius, a Northern European digital bank, deployed Mistral Large to upgrade its customer chatbot from a scripted platform to a generative AI solution. The new chatbot, available in Dutch, French, and…

CMA CGM and Mistral.

Mistral announced a strategic partnership with CMA CGM to deploy AI solutions across shipping, logistics, and media, aiming to streamline operations and enhance productivity. Use cases include cargo r…

Create and share a custom agent

Mistral introduced a feature allowing users to create custom agents in Le Chat with tailored instructions, tools, and knowledge bases. Users can define specific behaviors, enable tools like Canvas for…

Analyze data

Mistral AI introduced a new data analysis feature in Le Chat where users can upload spreadsheets and query them using plain language. The system automatically generates Python code (via pandas, matplo…

Activate Studio and generate an API key

Mistral introduced a free Experiment plan for its Studio product, allowing users to activate the service and generate an API key without a credit card. The plan provides access to Mistral's API for mo…

Test a model in the API playground

Mistral introduced a model playground in its Studio product, allowing users to interact with Mistral models via a chat interface, send prompts, and adjust generation parameters without coding. The pla…

Install Mistral Vibe and send your first prompt

Mistral released Mistral Vibe, a terminal-based tool for interacting with its AI models. Users can install and configure it via a one-liner command (macOS/Linux) or manual setup (macOS/Linux/Windows),…

Scaffold a project with Mistral Vibe

Mistral introduced Mistral Vibe, a tool that generates a full project structure from a natural-language description. Users can review and approve changes before Mistral Vibe applies them, ensuring con…

Send your first API request

Mistral AI published a quickstart guide for sending the first API request, requiring users to create an API key, install the SDK, and execute a chat completion request. The guide serves as a foundatio…

Build an agent with tools

Mistral introduced a new capability allowing models to call external tools mid-conversation, enabling access to APIs, databases, or internal services. The workflow involves defining a tool (e.g., a we…

Set up RAG with document search

Mistral introduced a quickstart guide for setting up Retrieval-Augmented Generation (RAG) with document search, enabling users to upload documents to a Library and query them in chat completions. The …

Draft a research report

Mistral AI introduced a new quickstart guide for drafting research reports using three core tools in Le Chat: Web Search, Document Uploads, and Canvas. The guide walks users through a 10-minute proces…

Bringing customers to the frontier.

Mistral AI has significantly expanded its customer base and strategic partnerships, highlighting collaborations with major corporations, governments, and institutions across industries such as maritim…

Ardian, a leading private investment firm managing $192 billion in assets, partnered with Artefact and Mistral AI to develop GAIA, an internal Generative AI platform that transforms how investment tea

Capgemini transforms software engineering with a self-hosted AI software engineering assistant powered by Mistral AI.

Capgemini, a global system integrator with 350,000 employees, launched a self-hosted AI software engineering assistant powered by Mistral AI’s Codestral model, transforming its software engineering ca…

Transforming the CX experience: How Cisco plans to cut renewal proposal time by 20%.

Cisco and Mistral AI partnered to develop an AI Renewals Agent that reduces renewal proposal creation time by 20% for Cisco’s Customer Experience (CX) team. The agent automates the analysis of over 50…

From PDFs to structured output: automated patent transcription at scale.

The European Patent Office (EPO) partnered with Mistral AI to deploy an advanced OCR system for automated patent transcription, achieving near-zero fault tolerance and processing 400,000 pages daily w…

Streamlining SAP S/4 HANA migration processes with intelligent locally hosted AI agent orchestration, powered by Mistral AI.

Mistral AI and SAP have partnered to integrate Mistral’s large language models into SAP’s enterprise AI solutions, enabling locally hosted, multilingual AI agent orchestration for SAP S/4 HANA migrati…

Beyond basic queries: Achieving 2x greater accuracy with text-to-sql on Snowflake.

Snowflake integrated Mistral AI’s LLMs (Mistral Large and Codestral) into its Cortex Analyst AI-powered assistant to enhance text-to-SQL capabilities, achieving nearly 2x greater accuracy than single-…

Building a more intelligent car: How Stellantis is accelerating automotive innovation.

Stellantis and Mistral AI have partnered to integrate Mistral’s large language models (LLMs) into automotive innovation, focusing on vehicle engineering, in-car experiences, and manufacturing. The col…