Engineers on the AI SDK can generate text or image to video with native audio by setting one model ID, with clips up to 15 seconds at 1080p, routed through AI Gateway's usage tracking and failover.
Key takeaways · AI-distilled
Grok Imagine Video 1.5 Lite from xAI generates 1 to 15 second clips at 480p, 720p or 1080p with native audio, in seven aspect ratios spanning widescreen, square and portrait.
In the AI SDK it is called through experimental_generateVideo with the model id spacexai/grok-imagine-video-1.5-lite plus resolution, aspectRatio and duration parameters. The AI CLI supports it too.
Vercel suggests chaining an image model with it to create a still and animate it in one flow. Calls work with an AI Gateway API key or your own provider key.
