Microsoft unveils MAI-Image-2.6-Flash with twice the speed of leading AI image models

Microsoft announced on September 4, 2026, the release of MAI-Image-2.6-Flash, a self-developed image generation model available via the Microsoft Foundry platform. The model is priced per 1 million tokens and is optimized for speed and cost efficiency while retaining core capabilities such as text-to-image and image-to-image editing. Microsoft recommends MAI-Image-2.6 for high-quality commercial use, while MAI-Image-2.6-Flash is suited for tasks prioritizing speed and cost. Mustafa Suleyman, Microsoft’s AI chief, stated that MAI-Image-2.6-Flash generates images twice as fast as GPT-Image-2-Medium, with a 72% improvement in GPU efficiency. The model supports both image creation and editing, offering features like object removal, text updates, and localized re-painting. The release follows Microsoft’s recent advancement in the Arena text-to-image ranking, where MAI-Image-2.6 secured second place.