Microsoft's image models land on AI Gateway: one key for text-to-image, and five reference images for product and layout edits
Synopsis
Microsoft's MAI-Image-2.6 and MAI-Image-2.6-Flash are now available on AI Gateway, letting developers generate and edit images through the AI SDK or AI CLI with one API key and no provider account. Both models support photorealistic imagery, text rendering, and targeted edits to products, portraits, and layouts, accept up to five reference images to guide edits or combine subjects, and output up to 1536 × 1536. MAI-Image-2.6 targets final assets where quality matters more than speed, while Flash targets faster, lower-cost generation for many iterations.
Interpretation
Developers can now call Microsoft's MAI-Image-2.6 and MAI-Image-2.6-Flash image models directly on AI Gateway, generating images from text or editing existing ones. Using such models previously required a separate provider account; now a single API key through the AI SDK or AI CLI provides access without a provider account. Both models are offered under the model IDs microsoft/mai-image-2.6 and microsoft/mai-image-2.6-flash, with generateImage call examples.
Both models support photorealistic imagery, text rendering, and targeted edits to products, portraits, and layouts. Edits can take up to five reference images to guide changes or combine subjects, with output sizes up to 1536 × 1536. The call example passes an image as input with an instruction, such as changing a vase to terracotta while keeping the flowers and setting.
The two models split by use: 2.6 for final assets when image quality matters more than speed, and Flash for faster, lower-cost generation when many iterations are needed, such as exploring prompts or generating variations. One interface offers both trade-offs, and users pick the model ID according to quality versus iteration count. The model IDs are microsoft/mai-image-2.6 and microsoft/mai-image-2.6-flash, and both can be tried in the model playground.
Perspective
This release is aimed at developers using AI Gateway and covers two call types: text-to-image generation and image editing, where edits can supply up to five reference images and outputs reach 1536 × 1536. The two models serve quality-first final assets and fast generation for many iterations respectively, and the model IDs and generateImage call form are given.
The release note gives no comparison against other image models and no quantitative figures for generation quality, latency, or cost; the descriptions of photorealism, text rendering, and targeted edits come without evaluation data, so actual results need to be verified with one's own prompts and reference images.
