ByteDance's Flagship Video Model

Seedance 2.5
Native 4K & Audio.

Generate native 30-second clips in 4K with synchronized audio and support for up to 50 multimodal references.

Trusted by creators at leading brands

Facebook
NVIDIA
Netflix
Google
Facebook
NVIDIA
Netflix
Google
Facebook
NVIDIA
Netflix
Google
Facebook
NVIDIA
Netflix
Google

30 seconds. 4K resolution. Co-generated audio.

Cinematic Motion

Ballet Dancer

Close-up on her gnarled, arthritic feet slowly sliding into first position on the dusty stage floor...

Convenience Store

A young foreign man and woman unexpectedly meeting inside a small convenience store late at night...

Running Bride

The bride slows down, turns back as if someone is calling her name, then laughs through tears...

Native 4K Output

Skip the upscaler. Seedance 2.5 generates professional 4K footage natively.

50+ References

Upload up to 50 multimodal references including images, video, and audio to guide your generation.

Native 4K Output

Skip the upscaler. Seedance 2.5 generates professional 4K footage natively.

50+ References

Upload up to 50 multimodal references including images, video, and audio to guide your generation.

Frequently Asked Questions

Seedance 2.5 is ByteDance's next-generation AI video model, generating native 30-second clips in 4K with synchronized audio and support for up to 50 multimodal references.
Seedance 2.5 extends clip length from 15 to 30 seconds, raises resolution from 1080p to native 4K, increases reference inputs from about 12 to 50, and adds local region-redraw editing.
Yes. Audio is co-generated within the same pass as the video, producing native synchronization between on-screen action and sound.
You can upload up to 50 multimodal references in a single generation. This includes images, video clips, audio files, and 3D white-models.
Seedance 2.5 supports native 4K resolution output combined with 10-bit color depth.