AI CURATED HOT STORY · 10/27/2026
Microsoft has released the MAI-Image-2.7-Preview model, which has rapidly ascended to the global top tier in image editing benchmarks. According to Artificial Analysis, it ranks second on the AA-Image-Editing v2.1 leaderboard, trailing only OpenAI's GPT Image 2.5. On Arena.ai's Image Edit Arena, the model secured third place with a score of 1470 points, a 45-point improvement over its predecessor, elevating Microsoft's institutional ranking to second.
Technically, MAI-Image-2.7 outperforms the previous version across all seven editing actions, with significant gains in composition, text rendering, and style transfer (Elo increases of ~48-57). It leads in specific tasks such as commercial design, portrait editing, and reasoning-based edits, demonstrating superior multimodal understanding. Currently available on MAI Playground for editing tasks, text-to-image capabilities are expected soon.
This release signals Microsoft's strong resurgence in generative visual AI, narrowing the gap with leaders like OpenAI. By offering high-performance, controllable alternatives, MAI-Image-2.7 is poised to accelerate the adoption of AIGC tools in advertising, design, and content creation workflows.
The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.