Microsoft AI CEO Mustafa Suleyman has announced a family of seven new MAI models spanning image generation, voice synthesis, transcription, and code generation. The lineup includes MAI-Thinking-1, a reasoning model with 35 billion active parameters and a 128K context window — trained entirely without distillation — alongside MAI-Code-1-Flash, optimized for GitHub and VS Code workflows. Other releases include MAI-Image-2.5 for image generation and MAI-Voice-2, which delivers expressive speech across 10 languages.

Building a Hillclimbing Machine: Launching Seven New MAI Models