Microsoft opens MAI-Image-2.6 in Foundry preview and adds a Flash variant

Microsoft has expanded developer access to its MAI image family. MAI-Image-2.6 and the new MAI-Image-2.6-Flash are now available in public preview through Microsoft Foundry, moving the higher-end model beyond its earlier limited-access phase and adding a lower-latency option for production workloads.\n\nBoth models support text-to-image generation and image editing. Microsoft also lists multi-reference editing, web grounding and dynamic aspect ratios among the current capabilities. Its Foundry documentation identifies both 2.6 models as preview deployments, which is the important availability fact for developers evaluating them now.\n\nThe Flash variant is positioned around throughput and cost. Microsoft says it generates faster and uses GPU resources more efficiently than GPT-Image-2-Medium. Those comparative figures are vendor claims rather than independent production benchmarks, so AiToolMap should treat them as context, not as direct rating evidence. Independent launch coverage confirms the public-preview expansion and the two-tier positioning.\n\nThe update matters because Microsoft is now offering its own current image models through a standard developer deployment surface rather than only showcasing them in model evaluations or playground access. Teams can choose the precision-oriented 2.6 model or the faster Flash version without changing away from the Foundry ecosystem.\n\nBefore either surface is cataloged or rated independently, AiToolMap should verify current token/output pricing, regional availability, safety and content-policy controls, reference-image handling and whether the two variants warrant separate listings or one model-family entry.