Microsoft AI has launched MAI-Image-2.6-Flash, an accelerated version of its flagship MAI-Image-2.6 model, and it's now available in public preview on Microsoft Foundry. The Flash model handles the same jobs as the original — text-to-image generation, image editing, multi-image references, fetching up-to-date information from the web, and automatic aspect ratio selection — with the focus on pushing latency and cost down further.
Speed is the headline feature. Microsoft says MAI-Image-2.6-Flash generates images at 2.8 times the speed of GPT-Image-2 Medium while being 72% more efficient with GPU resources.
But this isn't just a quality-for-speed trade-off. On Artificial Analysis' current image-editing leaderboard, MAI-Image-2.6-Flash ranks third with an Elo rating of 1311, just ahead of GPT Image 2 high's 1309. For text-to-image, it ranks eighth with an Elo rating of 1297. The flagship MAI-Image-2.6 remains stronger — it's currently first for image editing and second for text-to-image.
Microsoft Foundry lists image output for the Flash model at $19 per million tokens, versus $38 per million for the flagship 2.6. Converted to representative images, Artificial Analysis puts Flash at about $19.50 per 1,000 images, while GPT Image 2 high lands around $211 per 1,000 images.