MICROSOFT BREAKS FREE FROM OPENAI: SEVEN IN-HOUSE AI MODELS UNVEILED AT BUILD 2026, NONE OF THEM FROM SAM ALTMAN At Microsoft Build 2026 on June 2 and 3, the company that spent $13 billion turning OpenAI into the dominant commercial AI platform announced it no longer needs OpenAI for its frontline products. Microsoft unveiled the MAI model family: seven in-house AI models spanning coding, reasoning, image generation, voice, and transcription. The flagship for developers is MAI-Code-1-Flash, an inference-efficient coding model tuned for GitHub. It is already rolling out in VS Code through the GitHub Copilot model picker and sits above Claude Haiku 4.5 on price-to-performance for fast, low-cost coding tasks. Full public preview for GitHub Copilot is expected in September 2026. All seven MAI models will be available via API on Azure AI Studio with fine-tuning support. Microsoft says the family was built under what the company calls a Humanist Superintelligence philosophy, AI designed to serve people rather than replace them. It is the kind of framing that sounds good in keynotes and makes it harder for regulators to ask hard questions. The practical message is less gentle. Microsoft is telling the world it can build frontier AI by itself. The company has poured billions into OpenAI and watched it file for its own IPO while developing competing products. The MAI launch is the answer to that: we can do this ourselves, and we are doing it inside the tools a billion people already use every day. The OpenAI partnership is not over. But it just got a great deal more complicated for both sides. Keywords: Microsoft MAI models, Build 2026 AI, MAI-Code GitHub Copilot, Microsoft OpenAI independence 2026