xAI has released grok-imagine-video-1.5-preview, a new image-to-video model now available through the xAI API in preview, according to the company's release notes. The model generates video from still images using natural-language motion prompts, supports output up to 720p resolution, and preserves the lighting and visual style of the source image. xAI describes the model as designed to produce cinematic motion from a single frame. The preview release gives developers early access ahead of a potential general launch, and follows xAI's recent release of Grok Build, a terminal-based agentic coding tool. The move puts xAI in more direct competition with video generation models from Google, Runway, and others, at a time when image-to-video capabilities are becoming an expected feature from major AI platforms.