Microsoft just quietly trained an AI model that matches Claude Opus 4.6 — without a single token of OpenAI data. MAI-Thinking-1 hits 97% on AIME 2025 and a 256K context. MAI-Code-1-Flash beats Claude Haiku 4.5 by 16 points on SWE-Bench Pro and solves the same tasks with 60% fewer tokens. Translation: the OpenAI dependency Microsoft built its whole AI stack on is now optional. Track every model that ships with www.contentbuffer.com. Which frontier lab loses the most if Microsoft fully cuts the cord? #AI #Microsoft #LLM #DevTools #OpenAI