MAI-Image-2.7-Preview

MAI-Image-2.7-Preview is an advanced image-editing artificial intelligence model developed by Microsoft. Released initially in preview focusing specifically on editing capabilities rather than fresh text-to-image generation, the model is built to handle complex photo modifications through natural language prompts. It provides robust tools for scene restyling and relighting, asset restoration, and precise integrated text or symbol manipulation.

Core Capabilities and Performance

The model introduces improvements across all standard use cases and evaluation metrics compared to its predecessor, MAI-Image-2.6. It excels in reasoning-based edits that require an understanding of spatial relationships between objects, lighting, and background surroundings. According to independent benchmarks such as Artificial Analysis, MAI-Image-2.7-Preview ranks near the top of the category frontier for image editing, demonstrating strong performance in enhancement, restoration, and stylistic adjustments.

Prompt Guide and Tips

When interacting with MAI-Image-2.7-Preview, users achieve optimal results by providing clear instructions in English that define explicit parameters. Helpful prompting strategies include specifying the subject action (such as modifying or moving an element), detailing the target artistic style or lighting condition, and noting structural preservation constraints to ensure non-target areas remain intact. Specificity regarding environmental context helps align complex creative or professional edits with expected outcomes.