Microsoft Launches MAI-Thinking-1, Its First In-House Frontier Reasoning Model

Microsoft unveiled MAI-Thinking-1 at Build 2026, its first internally developed frontier reasoning model and the company’s most capable in-house model to date. The announcement marks a significant shift — Microsoft previously relied on OpenAI’s models and began developing its own only last year. The move follows a recent renegotiation of the Microsoft-OpenAI partnership that loosened exclusivity terms. No public benchmarks or API availability timeline were announced. MAI-Thinking-1 positions Microsoft as a full-stack AI lab rather than a model reseller.