Update: MAI Image 2.7 leaves preview at No. 1 in AI image editing and gains a Flash version

Microsoft launched on Tuesday (17), during Ignite 2026, the final version of MAI Image 2.7, its image generation and editing model, alongside a faster and cheaper variant, MAI Image 2.7 Flash. According to Microsoft AI, the model takes first place in image editing on the Artificial Analysis ranking, outperforming OpenAI's GPT Image 2.6 Sunburst (max) at about one fifth of the price.

On October 27, Geekiki had already covered the MAI Image 2.7 preview, then limited to image editing and ranked second on the same leaderboard. The final version extends the model to text-to-image generation and adds the Flash variant, aimed at high-volume use.

What changes in the final version

MAI Image 2.7 lets users change an object's color, material or position without disrupting the rest of the image, preserving visual consistency across successive edits, according to Microsoft. Both models also generate transparent backgrounds, useful for product images, logos, icons and illustrations without manual cutouts. In text-to-image generation, the model ranks fourth on Artificial Analysis — Microsoft presents itself as the second-best lab in the ranking. The model card lists 20 billion non-embedding parameters, a 32,000-token context window and a maximum output of about 4.2 megapixels.

Pricing, availability and Canvas

Both models are in public preview on Microsoft Foundry, priced at about $35 per 1,000 images for MAI Image 2.7 and about $10 per 1,000 images for Flash. They can also be tried in MAI Labs and through partners such as fal, Genspark, OpenRouter and Vercel. Microsoft also launched Canvas, an interactive workspace in MAI Labs powered by the Flash model, where users can start from a prompt or sketch, turn rough ideas into diagrams, and select any part of an image to replace, move or restyle. The MAI Image family is already the default model behind Bing Image Creator, with usage growing at a compounded weekly rate of about 15% since June, according to the company. Adobe, Figma Weave, Minute (Sports Illustrated), Wirestock and Manus are already working with MAI Image, Microsoft says.

The launch is part of the seven-model MAI family unveiled at Ignite 2026, which also includes the coding and security models already covered by Geekiki in Microsoft lança MAI Code 2 e MAI Cyber 2 Flash no Ignite 2026, plus real-time voice, transcription and model routing.

One caveat: the first-place ranking is published by Microsoft itself, using Artificial Analysis data from November 16 — 1,215 Elo points across 5,478 votes, just 4 points ahead of GPT Image 2.6 Sunburst (max) at 1,211. The lead is narrow, and the other performance figures cited by the company are its own measurements, with no independent verification so far.

Image: official artwork from the MAI Image 2.7 announcement, by Microsoft AI.

Fontes

Transparency note: this article was produced with the assistance of artificial intelligence, based on cross-checking public posts on X and sources on the internet, and reviewed against the original sources cited above. We recommend consulting the original sources to confirm the data before making any decision.

Por Muse