ByteDance launches Seedance 2.5 AI video model with longer clips and multimodal editing

ByteDance's Seed team announced the release of Seedance 2.5, the latest generation of its AI video creation model. Building on the multimodal audio‑video architecture of Seedance 2.0, the new version focuses on foundational generation and reference‑based generation, enabling up to 30‑second video clips in a single pass and supporting multiple rounds of extension for longer content.

Seedance 2.5 allows creators to input up to 30 images, 10 video clips and 10 audio clips as references, offering timestamp‑level control for precise editing, green‑screen effects, camera perspective changes and 4K output. The model is now available on the Jimeng AI and Doubao platforms, targeting brands, creators and production teams that need reliable drafts for campaigns, teasers, tutorials and story‑driven videos.