Foundry Labs @ Build 2026
Summary
Microsoft unveiled a refreshed Foundry Labs experience at Build 2026 on June 03, 2026, establishing labs.ai.azure.com as a central hub for cutting-edge AI experiments. The updated platform features a searchable catalog, "Stories" showcasing community projects, and "Communities" for events. Foundry Labs is now organized into six real-world AI impact areas: Biomedical Sciences, Chemistry & Materials, Code & Software Engineering, Creative & Generative Media, Geospatial & Earth Science, and Robotics & Physical AI, with examples like GigaTIME, MatterGen, BugPilot, MAI-Image-2, EO/OS Object Detection, and Rho-Alpha. This categorization also applies to the Microsoft Foundry Model Catalog. New MAI models are available, including MAI-Thinking-1 for reasoning, MAI-Image-2.5 (ranked No. 3 on Arena.ai) and MAI-Image-2.5 Flash for image generation, MAI-Voice-2 and MAI-Voice-2 Flash for multilingual text-to-speech, and MAI-Transcribe-1.5 for speech-to-text. These models power Microsoft products like Copilot and Azure Speech.
Key takeaway
For AI Engineers evaluating new models or platforms, Foundry Labs offers a centralized, categorized hub for Microsoft's latest AI experiments. You should explore labs.ai.azure.com to discover frontier-quality MAI models like MAI-Thinking-1 or MAI-Image-2.5, which are now available for building. Consider integrating these cost-efficient models into your applications, especially if you need specific modalities like advanced reasoning, image generation, or multilingual speech. Engage with the Foundry Discord community to influence future AI development.
Key insights
Microsoft's Foundry Labs centralizes AI experiments, offering new models and community features across six key impact domains.
Principles
- AI experiments benefit from centralized access.
- Community engagement drives AI adoption.
- Categorize AI by real-world problem.
Method
The article describes the reorganization of Foundry Labs around six AI impact areas and the release of new MAI models across four modalities (text, image, voice, speech) for developers.
In practice
- Explore labs.ai.azure.com for prototypes.
- Use MAI models for specific modalities.
- Engage with the Foundry Discord community.
Topics
- Foundry Labs
- MAI Models
- AI Experimentation
- Multimodal AI
- Generative AI
- Developer Community
- Microsoft Azure
Best for: NLP Engineer, Computer Vision Engineer, AI Scientist, AI Engineer, Machine Learning Engineer, Research Scientist
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by Microsoft Foundry Blog articles.