Build capable AI agents with memory & tools
- We’re committed to empower the AI community with open technology. Our open models sets the bar for efficiency, and are available for free, with fully permissive license. - Our optimized commercial models are designed for performance and are available via our flexible deployment options.
Enterprise APIs include regional data processing controls, system-level SLAs, SAML SSO, audit logs, and options for self-hosted deployment (private cloud, on-premises). EU-hosted servers and white-label capabilities support compliance-heavy sectors (finance, government, manufacturing).
Mistral Agents API enables creation of agentic workflows through Studio with agent orchestration, task scheduling, 100+ connectors (Notion, Slack, Linear, Drive), and persistent memory/skills. Users can deploy agents that autonomously execute complex work across integrated tools.
Mistral provides multiple API endpoints (Mistral Large 3, Medium 3.5, Small 4) with transparent per-token pricing, batch processing discounts (50% off), and detailed usage tracking via Studio dashboard. Developers can select models by performance/cost tradeoff and monitor spending in real-time.
Vibe for Code (powered by Devstral) provides CLI, IDE (VS Code/JetBrains), and remote agent capabilities for autonomous code generation, legacy translation, CI/CD automation, and test authoring. Developers can invoke agents directly in their workflow with context-aware reasoning.
Forge enables enterprise domain adaptation, end-to-end model training, reinforcement learning, and synthetic data generation. Open-weight models (Apache 2.0 for research; commercial license for production) allow self-hosting and derivative customization with full control.