Microsoft’s Bold Move: Unveiling Seven Homegrown AI Models for Long-Term Self-Sufficiency

Microsoft showcased a groundbreaking move at the Build developer conference by introducing seven new in-house AI models developed by the Microsoft AI Superintelligence Team. This strategic initiative aims to reduce reliance on external partners like OpenAI and Anthropic.

Mustafa Suleyman, CEO of Microsoft AI, emphasized the importance of building trustworthy in-house models for long-term self-sufficiency and credibility in the AI industry.

Despite being a major investor in OpenAI and Anthropic, Microsoft recognizes the need to assert more control over its AI future, especially with the increasing partnerships between its competitors and these external AI providers.

The highlight of the new MAI models is MAI-Thinking-1, a reasoning model that Microsoft claims performs on par with Anthropic’s Claude Sonnet 4.6 in human testing and matches Claude Opus 4.6 on a popular coding benchmark.

MAI-Thinking-1 was developed without relying on distillation from other companies’ models, making it appealing to enterprises concerned about data lineage transparency.

Microsoft has made MAI-Thinking-1 available for private preview on Microsoft Foundry, alongside models from OpenAI and Anthropic.

In addition to MAI-Thinking-1, Microsoft also introduced MAI-Code-1-Flash, a coding model with 5 billion parameters that is being integrated into Visual Studio Code and GitHub Copilot. Another model, MAI-Image-2.5, ranks high on an image-editing leaderboard, surpassing Google’s Nano Banana Pro.

The complete range of models covers various domains such as image processing, voice recognition, transcription, coding, and reasoning.