Microsoft AI bets on cheap specialist models instead of chasing the frontier

Microsoft AI says it will prioritize low-cost specialist models and model orchestration over chasing the strongest general-purpose frontier model.

Its cybersecurity model MAI-Cyber-1-Flash reportedly beats Anthropic’s Mythos on CyberGym at roughly half the cost.

MAI-Image-2.5-Flash is said to use up to 84% less GPU compute than GPT-Image-2 for image generation.

Microsoft is also building routing systems that can switch between cheaper specialists and larger reasoning models as needed.

The shift reflects an industry move from raw model scale toward efficiency, speed, flexibility, and lower commercial costs.