xAI: Grok Imagine Video 1.5 Generations API

Grok-Imagine-Video 1.5 is xAI's current flagship video model, officially named `grok-imagine-video-1.5`. Over 1.0 it adds native 1080p output and extends reference-conditioned generation to audio — alongside reference images you can supply up to 3 preset voices so characters speak in a chosen voice. It supports text-to-video, image-to-video (first frame) and reference-to-video, up to 15 seconds per clip, with synchronized audio.