Mistral Document AI (with OCR 4) and Mistral Medium 3.5 arrive in Microsoft Foundry

2026-06-24 · Source: Microsoft Foundry Blog articles · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure · Depth: Intermediate, medium

Summary

Microsoft Foundry is expanding its AI model catalog with two new Mistral offerings: Mistral Document AI (with OCR 4) and Mistral Medium 3.5. Arriving today, Mistral Document AI with OCR 4 provides structured document understanding, featuring paragraph-level bounding boxes, block classification, and support for 170 languages. It is priced at \$4 per 1,000 pages for OCR only or \$5 with annotations. Mistral Medium 3.5, available tomorrow, is a 128B parameter dense model with a 256K token context window, designed for reasoning, coding, and agentic workloads. This open-weight model, released under a modified MIT license, supports both text and image inputs and costs \$1.5 per 1M input tokens and \$7.5 per 1M output tokens. These additions enable developers to build comprehensive AI systems within Foundry, combining document processing with advanced reasoning capabilities.

Key takeaway

For AI Engineers building enterprise-scale document processing or agentic AI systems, the integration of Mistral Document AI (with OCR 4) and Mistral Medium 3.5 into Microsoft Foundry provides critical new capabilities. You can now reliably convert complex documents into structured, pipeline-ready data and then apply a 128B parameter open-weight model for advanced reasoning and coding, all within a governed platform. This enables more robust RAG systems and automation, ensuring compliance and scalability for your production workloads.

Key insights

Mistral's new models in Microsoft Foundry enable integrated, structured document processing and advanced reasoning for enterprise AI.

Principles

• Production AI systems require diverse model capabilities.
• Structured document outputs enhance downstream AI logic.
• Open-weight models offer customization and cost efficiency.

Method

Deploy Mistral models via Microsoft Foundry's model catalog using real-time endpoints or serverless API, then integrate into applications for document pipelines or copilots.

In practice

• Build RAG systems with citation-ready document chunks.
• Power agentic workflows acting on document elements.
• Create large-scale enterprise ingestion pipelines.

Topics

• Mistral Document AI
• OCR 4
• Mistral Medium 3.5
• Microsoft Foundry
• Structured Document Understanding
• AI Agents

Best for: AI Architect, Machine Learning Engineer, NLP Engineer, AI Engineer, MLOps Engineer, Director of AI/ML

Related on AIssential

• OCR Processor - Mistral AI — Mistral AI's OCR API extracts structured text from diverse document types, preserving formatting and providing confidence scores.
• Unlocking Document Understanding with Mistral Document AI in Microsoft Foundry — Mistral Document AI 2512 converts complex unstructured documents into structured data with high accuracy and multilingual support.
• Unlocking document understanding with Mistral Document AI in Microsoft Foundry — Mistral Document AI transforms unstructured documents into actionable, structured data using advanced OCR and intelligent understanding.
• Gemma 4 now available in Microsoft Foundry — Gemma 4, a multimodal open-weight model, is now available in Microsoft Foundry, offering extensive capabilities for enterprise AI.
• Mistral bets on ‘build-your-own AI’ as it takes on OpenAI, Anthropic in the enterprise — Mistral Forge is a new platform from the French AI startup Mistral, announced at Nvidia GTC, designed to enable enterprises and governments to build custom AI models trained from scratch on their prop…
• Mistral Small 4 in 8 mins! — Mistral Small 4 unifies reasoning, coding, and multimodal capabilities for enterprise applications under an Apache 2.0 license.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Microsoft Foundry Blog articles.