According to Intel Beating monitoring, several users on X have posted that the Grok web version is undergoing a soft launch of the Imagine Agent Mode (beta). This is an additional agent mode on top of the existing Grok Imagine image-to-video generation, where the interactive interface has switched from a traditional chat window to an infinite canvas.
User screenshots show that Grok has added an "Imagine" entry in the sidebar, which leads to an open canvas. On the right, preset workflow templates are listed: Create Worlds, Short Film, UGC Product Stories, Brand Identity. Users can describe in one sentence what they want, and the agent progressively generates all the images and videos on the canvas. Specific capabilities include simultaneous generation and editing of multiple images, automatic conversion of static images to videos with auto-stitching, video trimming and transitions, and export of the final product.
TestingCatalog states that the Imagine Agent can handle complex multi-step instructions, such as "generate a 1-minute short film," "generate a whole set of comics," "generate a UGC product image-text package." Another user demonstrated the process of the agent first batch-generating 3 sets of product images and 3 sets of model photos, and then automatically combining them into a social media content package.
xAI has not yet made an official announcement.
