Grok Imagine 1.5, xAI's image-to-video, is entering preview on the API.

xAI launches its grok-imagine-video-1.5-preview image-to-video model on its API, generating 720p clips with native audio to lead the LMSYS Arena.

A still image as input, a prompt describing the movement, and the scene comes to life. The grok-imagine-video-1.5-preview model, now accessible via xAI's API, converts any starting frame into a fluid and cinematic video: camera movements, rhythm, and atmosphere are directed in natural language, down to the sound design, with audio being natively generated in a single pass (music, ambient sounds, synchronized dialogue).

Faithful to the starting frame, the model preserves the detail and lighting of the source image and extends the scene rather than reinterpreting it, allowing for shots to be chained into long sequences with a consistent aesthetic. Clips go up to 720p for durations ranging from six to fifteen seconds. On the 720p Image-to-Video Arena, this version holds the top spot ahead of Seedance 2.0, with a gain of 52 Elo points over the previous one. As for pricing, generation costs approximately 0.08 dollars per second of 720p video, plus 0.01 dollars per input image, and the video mode is also set to join the Grok interface via Imagine Agent Mode, without requiring code.