Microsoft's Lean AI Model Ditches OpenAI Dependency

Microsoft released MAI-Image-2-Efficient, a streamlined image generation model promising four-times faster GPU throughput and 41% cheaper output pricing than its flagship MAI-Image-2. The model outpaces Google's Gemini 3.1 Flash in latency and targets high-volume tasks like UI mockups and marketing assets. The release accelerates Microsoft's push to reduce reliance on OpenAI, which it added to its competitors list in 2024. MAI models now replace OpenAI's DALL-E in Copilot, letting Microsoft retain revenue previously paid as licensing fees.