Woofun AI reports that Microsoft has integrated specialized language models into its productivity suite, deploying an unnamed model within Excel and MAI-Code-1-Flash in GitHub Copilot. The Excel model delivers performance comparable to GPT-5.6 on standard tasks while maintaining a smaller footprint suitable for H100 and A100 hardware, resulting in reduced operational expenses.

MAI-Code-1-Flash, already adopted by millions of developers, achieves a code acceptance rate approximately 10% higher than GPT-5.4 Mini and Claude Haiku 4.5, alongside a 10% reduction in token consumption. Microsoft is applying this same training methodology to develop specialized models for Copilot Chat, Outlook, and PowerPoint.