Microsoft launches seven new MAI models with Frontier Tuning, enabling custom AI training on organizational workflows

Microsoft announced seven new MAI (Microsoft AI) models available on Foundry, OpenRouter, Fireworks, and Baseten, built from scratch with zero distillation and sharing the same data discipline, infrastructure, and evaluation framework.

Frontier Tuning allows organizations to train models directly on their own workflows using reinforcement learning environments (RLEs), keeping institutional knowledge and data proprietary; a MAI model tuned for Excel matched GPT 5.4 performance while being up to 10× more efficient, and when tuned for McKinsey's standards, MAI achieved the highest win rate of any model tested at roughly 10× lower cost.

Microsoft and Mayo Clinic are co-creating a frontier AI model for healthcare combining Mayo Clinic's clinical expertise and de-identified data with Microsoft's foundational AI capabilities; the model will first be deployed within Mayo Clinic, then made available to other organizations via Azure Foundry.

Microsoft reports a 1.4x efficiency boost from co-designing with its Maia 200 silicon, and commits to publishing in-depth safety and technical reports for transparency.