ByteDance launches Seedance 2.5, extending single-pass AI video generation to 30 seconds with audio
AI video model releases
ByteDance launches Seedance 2.5, extending single-pass AI video generation to 30 seconds with audio
Seedance 2.5 generates up to 30 seconds of audio-video in a single pass (up from 15 seconds in Seedance 2.0), and supports multi-round extensions for multi-minute outputs.
The model accepts multimodal reference inputs in one pass: up to 30 images, 10 video clips, and 10 audio clips, improving clay render, motion, and creative referencing.
New editing features include timestamp-level control for targeted audio and video edits, plus green screen, camera perspective, and reference-based editing aimed at film and advertising workflows.
Seedance 2.5 is rolling out now on Jimeng AI and Doubao Pro, with API access coming soon via BytePlus ModelArk.
ByteDance built the model on the unified multimodal audio-video joint-generation architecture introduced with Seedance 2.0, focusing on foundational and reference-based generation.
cho.sh