xAI ships Grok Imagine Image 2.0 with precise image editing

On 9 August 2026, xAI released Grok Imagine Image 2.0, the next generation of its image model inside the Grok assistant. The focus is less on raw image generation than on control: users can change specific parts of an image instead of rolling a completely new one each time.

New is precise partial editing – for example swapping an object or adjusting a region while the rest of the image stays stable. Add to that support for multiple input images: up to five references can be combined to steer style, subject and composition deliberately.

By its own account, the model ranks second on global arena leaderboards for text-to-image and image editing. Such community rankings should be taken with a grain of salt but give a sense of competitive position. A public API was still missing at launch.

For content production and marketing, the trend is decisive: image AI is evolving from a random generator into a controllable tool. Exactly this controllability – consistent subjects, targeted corrections, reusable references – decides whether AI images hold up in professional workflows. AI IN LIFE, too, uses consistent image and avatar references so our reports look visually of one piece.