Microsoft dropped seven new AI models in a single day at the Build developer conference in San Francisco. The lineup, announced June 2, marks the company’s largest single expansion of its in-house AI capabilities. The models span reasoning, image generation, coding, voice, and transcription-all under the Microsoft AI brand, built by the team led by CEO Mustafa Suleyman.
The flagship model, MAI-Thinking-1, is Microsoft's first dedicated reasoning model, designed to handle multi-step logic before responding. MAI-Image-2.5 delivers image generation and editing, while MAI-Code-1-Flash targets fast, enterprise-ready coding tasks. All seven are built for customization through Microsoft Foundry.
This rapid rollout follows three models shipped in April and signals a strategic shift. For years, Microsoft relied on OpenAI for AI innovation. Now, with its own models, the company gains full control over its roadmap and avoids negotiating API terms with a third party. The timing is critical as Microsoft and OpenAI reportedly renegotiate their partnership-each MAI release shifts the leverage.
The models are designed to integrate deeply into Copilot, Microsoft 365, GitHub, and Azure, allowing Redmond to optimize the full stack independently. Enterprise customers stand to benefit from seamless integration across the Microsoft ecosystem.