AI Models

Native 4K at 60fps, up to 15 seconds, multi-shot sequencing with 6 cuts, integrated audio generation.

Unified multimodal model. Text, image, and video input. Reference-based generation, video editing, and character consistency.

Transfer motion from reference videos to static images. Dance, gestures, choreography with identity preservation.

Edit videos with natural language. Describe changes in text while preserving camera paths, motion, and timing.

Turn portraits into talking videos with audio-synchronized lip-sync. 1080p at 48fps. Supports photos, cartoons, and illustrations.

Cinematic motion with native audio and lip-sync. Text-to-video and image-to-video, 1080p, up to 10 seconds.

Control element movement with motion paths and reference videos. Up to 6 independent elements with static brush support.

Ultra-fast 6B parameter text-to-image model with photorealistic quality and bilingual text rendering.