The largest single expansion of Microsoft's in-house AI lineup spans reasoning, coding, image editing, and more, signaling a clear pivot away from OpenAI dependence.
Microsoft just dropped seven AI models in a single day. Announced on June 2 at the Build developer conference in San Francisco, the new models represent the company’s largest single expansion of its in-house AI lineup to date. They cover reasoning, image generation and editing, coding, voice, and transcription, all built under the Microsoft AI (MAI) brand by the team that CEO Mustafa Suleyman assembled specifically for this purpose.
What Microsoft actually shipped #
MAI-Thinking-1 is the company’s first dedicated reasoning model, a system that works through multi-step logic problems before responding.
MAI-Image-2.5 handles image generation and editing, aimed at businesses that need visual content creation integrated into their existing Microsoft toolchains.
MAI-Code-1-Flash targets coding tasks, with the “Flash” designation hinting at speed optimization.
All seven models are designed for enterprise customization, meaning businesses can fine-tune them for specific use cases. Delivery will run through Microsoft Foundry and related platforms.
The Suleyman factor and the OpenAI question #
The MAI Superintelligence team responsible for these models was formed in November 2025 under Mustafa Suleyman, a co-founder of Inflection AI, who Microsoft brought on as CEO of its AI division. Three models shipped in April 2026, and those are already available for commercial use through Foundry. Now seven more in June. That’s ten in-house models in roughly two months.
Microsoft’s AI strategy for years essentially amounted to funding OpenAI, integrating GPT models into products, and collecting Azure revenue. Building proprietary models gives Microsoft full control over the roadmap for enterprise customization without negotiating API terms with a third party.
What this means for the enterprise AI market #
The models are intended for integration into existing Microsoft products like Copilot, which is already embedded across Microsoft 365, GitHub, and Azure, allowing Microsoft to optimize the full stack without coordinating releases with an external partner.
The two companies are reportedly renegotiating terms of their partnership, and every new MAI model release shifts the leverage equation. Microsoft has been ramping up proprietary AI development under the MAI label since mid-2025, beginning with MAI-1-preview in August 2025 and MAI-Image-1 in October 2025.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our