Microsoft Build 2026 Introduces MAI Models for Code, Images, Transcription, and Speech

Microsoft announced MAI-Code-1 for GitHub Copilot and VS Code, MAI-Image-2.5 for generation and editing, MAI-Transcribe-1.5 for speech recognition in 43 languages, and MAI-Voice-2 for speech synthesis with more than 15 additional languages. Microsoft said transcription was up to five times faster than competitors and its image model outscored Google Nano Banana Pro on an Elo benchmark. Microsoft began integrating MAI-Image-2.5 into PowerPoint and plans OneDrive support. The models were among seven MAI announcements at Build 2026.

Visit original source