MAI-Image is a family of image generation models made by Microsoft AI, part of Microsoft. A text-to-image model creates a picture from a written description (a prompt); the newer versions also edit an existing image according to an instruction. The versions offered on Microsoft Foundry are MAI-Image-2.5, 2.5-Flash, 2.5-Pro, 2.6 and 2.6-Flash; MAI-Image-2.7, an image editing model, entered preview on 27 October 2026.
MAI-Image-2.6 is a diffusion model: it starts from random noise and transforms it step by step into an image that matches the prompt, trained with a flow-matching loss. It accepts text and images as input, can use current web content and documents as context, and makes targeted edits, such as changing objects, layouts or text in an image or removing motion blur. MAI-Image-2.6-Flash costs less than half the price of the flagship model.
Developers deploy the models in Microsoft Foundry, Microsoft's Azure platform for AI models, and call an image generations API and an image edits API, billed by use on an Azure subscription. The MAI Playground lets people try them.