Microsoft MAI models — In-house reasoning and coding models, matching Opus 4.6, going into GitHub Copilot
Key Questions
What are Microsoft's new MAI models?
Microsoft unveiled MAI-Thinking-1 reasoning model and MAI-Code-1 coding model, both trained from scratch without distillation.
How does MAI-Thinking-1 compare to Claude?
MAI-Thinking-1 matches Claude Opus 4.6 performance on SWE-bench Pro.
Where will MAI-Code-1 be integrated?
MAI-Code-1 will be integrated into GitHub Copilot, reducing Microsoft's reliance on OpenAI and Anthropic models.
What does the MAI launch signal for Microsoft?
The release marks Microsoft's direct push into foundation models and a shift in the AI coding assistant competitive landscape.
How did the market react to the MAI announcement?
Microsoft stock dipped slightly following the announcement of its in-house AI models.
Microsoft unveils MAI-Thinking-1 reasoning and MAI-Code-1 coding models, trained from scratch without distillation. MAI-Thinking-1 matches Claude Opus 4.6 on SWE-bench Pro. MAI-Code-1 will be integrated into GitHub Copilot, reducing reliance on OpenAI and Anthropic. This shifts the competitive landscape for AI coding assistants.