ByteDance's Seedance 2.5 generates 30-second videos with 50 inputs
Up to 50 multimodal references in a single 30-second native video.
Deep Dive
ByteDance is reportedly launching Seedance 2.5 as early as this week, its next-generation AI video model that generates 30-second native videos with support for up to 50 multimodal references.
Key Points
- Seedance 2.5 generates up to 30 seconds of native video, doubling typical output lengths.
- Supports up to 50 multimodal references (text, images, video, audio) for rich input control.
- ByteDance targets launch as early as this week, competing with OpenAI Sora and Runway Gen-3.
Why It Matters
Longer, more controllable AI video generation for creators means faster, cheaper professional short-form content production.