Microsoft has dramatically expanded its AI capabilities at Build 2025, positioning Azure AI Foundry as a central hub for enterprise AI development and deployment.
The most notable addition is the integration of several powerful new models into the Azure AI Foundry Models catalog. Grok 3 from Elon Musk's xAI is now available, offering advanced reasoning, coding, and visual processing capabilities. This partnership combines xAI's cutting-edge models with Azure's enterprise-ready infrastructure, allowing developers to access Grok 3's capabilities in a secure, scalable environment.
Other significant model additions include Flux Pro 1.1 from Black Forest Labs (coming soon) and OpenAI's Sora video generation model (arriving soon in preview). Microsoft has also expanded its open-source offerings, with over 10,000 Hugging Face models now available in Foundry Models.
A key announcement is the general availability of the Model Router in Azure AI Foundry, which automatically selects the optimal Azure OpenAI model for different prompts. This intelligent routing system evaluates factors like query complexity, cost, and performance to deliver high-quality outputs while minimizing expenses. The router can choose between smaller, cheaper models for simpler tasks and larger, more powerful models for complex reasoning tasks.
Microsoft also unveiled Microsoft Discovery, an enterprise agentic platform designed to accelerate scientific research and development. This extensible platform enables scientists to collaborate with specialized AI agents to transform the entire discovery process—from knowledge reasoning and hypothesis formulation to experimental simulation and analysis. The system uses a graph-based knowledge engine to construct relationships between proprietary data and external scientific research, helping researchers drive scientific outcomes with greater speed and accuracy.
Additionally, Azure AI Foundry Agent Service has reached general availability with Multi-Agent Orchestration capabilities. This allows developers to orchestrate multiple specialized agents to handle complex tasks, with support for Agent-to-Agent (A2A) communication and the Model Context Protocol (MCP).
These developments highlight Microsoft's commitment to building a diverse AI ecosystem and establishing Azure as a comprehensive platform for enterprise AI deployment. With ChatGPT—built entirely on Azure—now serving over 500 million weekly active users, Microsoft has demonstrated the platform's ability to support AI applications at massive scale.