GitHub

Microsoft announced a number of new internal AI models at Build 2026, including a new "flagship" model: MAI-Thinking-1. This is an ambitious step in model development for Microsoft, which introduced its first internal models last year – previously it relied on OpenAI models. The two companies recently renegotiated their agreement to loosen ties.

Microsoft's first advanced reasoning AI is here

MAI-Thinking-1 is one of seven new models announced during Build 2026.

According to Microsoft, MAI-Thinking-1 is a “mid-sized model” that “matches leading models” on “key” software engineering criteria. Microsoft says the company "trained it on clean data from the start, without distilling third-party models."

As for the other models announced today, they focus on image generation, transcription, voice and coding.

MAI-Image 2.5 and the flash version can perform text-to-image conversion and image editing. MAI-Transcribe-1.5 is “five times faster than competing models”. MAI-Voice-2 and the flash version of that model (which Microsoft says is "coming soon") add 15 new languages and new options for voices. The new coding model, MAI-Code-1-Flash, is "inference efficient" and is integrated into GitHub Copilot and Visual Studio Code.