At Build 2026, Microsoft introduced MAI (Microsoft AI) — seven in-house models spanning reasoning, coding, image, voice, and transcription, marking a shift away from its heavy reliance on OpenAI.
The flagship, MAI-Thinking-1, is a 35-billion-parameter reasoning model with a 256,000-token context window, trained entirely from scratch with no distillation from other labs. In blind testing, it was preferred over Claude Sonnet 4.6 and matched Claude Opus 4.6 on the SWE-Bench Pro benchmark.
The rest of the family includes MAI-Code-1-Flash (already in VS Code and GitHub Copilot), MAI-Image-2.5 (live in PowerPoint, ranked No. 2 on Arena.ai), MAI-Voice-2 for natural speech generation, and MAI Transcribe 1.5, which supports 43 languages.
Microsoft is emphasizing cost efficiency — an Excel-tuned MAI model reportedly matches GPT-5.4 at roughly a tenth of the cost — and is distributing the models beyond its own products via OpenRouter, Fireworks, and Baseten. A Mayo Clinic partnership for a healthcare-specific model was also announced.
For the Microsoft 365 and Copilot ecosystem, MAI signals that Microsoft is now building its own AI foundation, not just distributing others'.
Watch the MS Build video here: Microsoft AI CEO unveils 7 new AI models | Mustafa Suleyman at Microsoft Build 2026
Read below articles to know more
Building a hill-climbing machine: Launching seven new MAI models
New MAI models in Microsoft Foundry across text, image, voice, and speech
No comments:
Post a Comment