grok-video-1.5-max (Grok Imagine Video 1.5 Max) Top-tier autoregressive multimodal video model released by xAI in 2026, built on Aurora MoE architecture. Up to 720p/15s, generates synchronized visuals + audio in one pass, with superior physical simulation and cross-frame character consistency. Faster inference than competitors, supports text/image-to-video and clip chaining, ideal for bulk short-video & ad creative production.