From the source
Microsoft made its Decision-1.1 decision model generally available and added new first-party MAI models to Foundry, its enterprise platform for grounding, testing and deploying AI applications and agents.
- Decision-1.1 is a fast, low-cost generative model that makes decisions dynamically instead of relying on hard-coded rules, and Microsoft says it offers superior decision-making to large language models.
- The company says LLMs waste time thinking through routing, classification, prioritization, verification and workflow control, while decision models reach answers quickly within constrained definitions and return calibrated confidence scores so applications can defer when uncertainty is high.
- The model builds on Microsoft-Decision-1, released in October, and Microsoft says it is built on a new base, Google's Gemma 4 26B A4, adding multimodal support for analyzing images in instructions.
- MAI-Code-2 and MAI-Code-2-Flash are now generally available on
GitHub, with Code-2-Flash in private preview in Foundry, while MAI-Image-2.7 and MAI-Image-2.7-Flash are in preview for sharper editing and more consistent image generation.
- MAI-Cyber-2-Flash is available in preview in Codename MDASH, Microsoft's agentic code security system in preview for eligible organizations.
- Foundry IQ now provides an enterprise knowledge graph out of the box, and Foundry uses transparency and observability capabilities to create learning loops, including human annotations in preview that let experts rate and comment on agent traces.