xAI

Grok Imagine generates images from prompts and can edit or combine source images using natural-language directions. The Quality variant adds 1K and 2K output choices and is the better fit when resolution matters, while the standard entry is suitable for ordinary iteration. For edits, keep each reference's role explicit so the model can distinguish subject, style, and composition sources.

The main Grok Imagine Video model can generate a clip from text, animate an image, incorporate reference images, edit a source video, or continue a video from its end. These modes use different inputs and constraints, while Grok Imagine Video 1.5 in Omnipix is the direct image-to-video path and should not be presented as a replacement for every workflow of the main model.