What is Seedance 2.0?

What changed with 2.0
Earlier Seedance versions took text and an image. 2.0 put audio and video generation into one architecture and let a request mix references: who from a picture, how from a clip, which voice from a recording. That made reference-driven video mainstream within days of launch, and drew complaints from film studios over clips of famous faces.
What it is good at
- Keeping a character recognisable when the face is given as an image.
- Speech with moving lips, and the sound of the room, in the same render.
- Camera moves named in the prompt: push-in, tracking, handheld.
Where it stops
Takes run up to 15 seconds on this site; the 2.5 generation doubled that to 30. Small lettering in the frame can garble, as on every current video model. Seedance 2.5 vs Seedance 2.0
Seedance 2.0 in SceneMixer
- Seedance 2.0: 1080P, 63 credits a second.
- Seedance 2.0 Fast: 720P, 25 credits a second.
- Seedance 2.0 · 480P: 480P, 13 credits a second.
Pick it from the model button in the episode editor’s top bar (it shows the current tier’s name, next to Composite), or in Settings → Model Preferences → Video Model. The choice applies to the segments you render next; segments already made keep the tier they were made with, and each version records it.
Sources
- ByteDance Seed blog, February 12, 2026 (Chinese) — the Seedance 2.0 launch.
Checked on November 16, 2026; SceneMixer’s figures are read from the live price list when the page is built.
FAQ
What is Seedance 2.0?
ByteDance's video model from February 2026 that takes text, images, audio and video as references.
Who made Seedance 2.0?
ByteDance's Seed team.
Does Seedance 2.0 have sound?
Yes. Speech, ambience and effects come with the picture.
How long can a Seedance 2.0 clip be?
Up to 15 seconds per take on this site.
Is Seedance 2.0 still worth using?
Yes for 15-second scenes; the 2.5 family is better for long takes.
Where can I use Seedance 2.0?
In ByteDance's apps and API, and in SceneMixer as three video tiers.
Updated November 16, 2026