ByteDance launches Seedance 2.5 — 30s one-take AI video
ByteDance announced Seedance 2.5 on 2026-07-31, an updated generative audio-video model that produces up to 30-second clips with synchronized audio in a single pass (double the previous 15s). The model accepts expanded multimodal references — up to 30 images, 10 video clips and 10 audio clips — and introduces multi-round extension to chain generations while preserving characters, scene and style. Seedance 2.5 adds professional editing controls (timestamp editing, green-screen, camera-perspective changes) and is available now inside ByteDance-operated Jimeng AI and Doubao Pro. Programmatic API access is announced as coming soon via BytePlus ModelArk; weights are not open-sourced and pricing/quotas have not been published.
- •ByteDance announced Seedance 2.5 on 2026-07-31.
- •Seedance 2.5 generates up to 30-second video clips with synchronized audio in a single pass (previously 15 seconds).
- •The model supports up to 30 images, 10 video clips and 10 audio clips as reference in a single generation.
