ByteDance Seed · Video model · Released July 31, 2026

Seedance 2.5 video generator

ByteDance’s Seedance 2.5 renders a single take of up to 30 seconds with sound and follows up to 50 references. On Flowith it sits on a canvas next to your characters, storyboard and editor, so a clip becomes a shot in a film.

“Fog on the rock”, 1080p, from a storyboard frame
Shot from a 3D Director Stage previs, 1080p
“Dawn”, 1080p, from The Keeper

Seedance 2.5 on Flowith

The settings you get in the Flowith composer. Checked on November 14, 2026.

Clip length
4 to 30 seconds per take, or Auto
Resolution
480p, 720p, 1080p
Aspect ratio
Adaptive, 16:9, 9:16, 1:1, 4:3, 3:4, 21:9
References
Up to 30 images, 10 video clips and 10 audio clips (50 in total)
Frames
Start frame and end frame
Sound
Generated with the picture
Characters
Saved characters can be @-mentioned into a shot

What Seedance 2.5 is good at

One take, twice as long

ByteDance doubled single-pass generation from 15 seconds in Seedance 2.0 to 30 seconds, with audio generated in the same pass.

Fifty references in one shot

Mix up to 30 images, 10 video clips and 10 audio clips. ByteDance also lists clay-render references, so a grey 3D layout can set blocking, camera and light.

Direct by the timestamp

ByteDance describes timestamp-level control: direct the story, camera and motion for a time range, then change characters or actions in one part of the clip.

Seedance 2.5 Flash

A faster, lower-cost edition with smaller limits on Flowith: up to 720p and 10 image, 5 video and 5 audio references. It shows as Coming soon in the model picker until ByteDance opens access.

How to use Seedance 2.5 on Flowith

  1. Step 1

    Open Flowith with Seedance 2.5 picked

    The button above opens a canvas with the model already selected in the composer.

  2. Step 2

    Describe the shot

    Say what happens, how the camera moves and what it sounds like. Add a start or end frame, reference images or clips, or a saved character.

  3. Step 3

    Set length, resolution and aspect ratio

    Generate, and the clip lands on the canvas as a card next to its references.

  4. Step 4

    Keep going

    Reshoot, continue from the last frame, or put the shots in order in the editor. Or ask Creator Agent to plan the film and pick a model for each shot.

Seedance 2.5 vs other video models on Flowith

All of these run on the same canvas, so you can try one prompt on several models and keep the best take.

ModelClip lengthResolutionReferencesPick it for
Seedance 2.5 4–30 s480p–1080p30 images, 10 videos, 10 audioLong single takes that follow many references
Kling 4.0 3–30 s720p, 1080p10 images, 5 videosPeople, motion and start/middle/end frames
Wan 3.0 5, 10, 15, 20, 30 s480p–1080pStart imageText or image to video with sound
Gemini Omni 1.2 Flash 3–15 s360p–4K6 images, 3 videosCheap drafts, then edits and upscales
Veo 3.1 4, 6, 8 s720p–4K, by variantStart imageShort clips with native sound

Seedance 2.5 questions

Seedance 2.5 is ByteDance Seed’s video model, released on July 31, 2026. It generates up to 30 seconds of video with synchronized audio in one pass and accepts up to 30 images, 10 video clips and 10 audio clips as references.
Each take can be 4 to 30 seconds, or Auto. For longer pieces, render several shots on the canvas and join them in the editor, or continue a shot from its last frame.
On Flowith you can choose 480p, 720p or 1080p. Draft at 480p while you work out the shot, then render the keeper at 1080p.
Yes. Add the character’s sheet or saved Flowith character as a reference in every shot. With up to 30 reference images you can include turnarounds, outfits and props together.
Flash is the faster, lower-cost edition. On Flowith it goes up to 720p and takes up to 10 images, 5 videos and 5 audio clips; Seedance 2.5 goes up to 1080p with 30, 10 and 10. If Flash still shows Coming soon in the picker, use Seedance 2.5.
Signing up is free. Each generation spends Flowith credits, and the cost depends on the settings you choose. Current rates for every model are on the pricing page.

Sources

The Keeper · film
The Keeper · film
Product photo
Product photo
The Keeper · film
The Keeper · film
Poster
Poster
Character · Fennec
Character · Fennec
3D illustration
3D illustration
Portrait
Portrait
The Keeper · film
The Keeper · film

Every frontier model. One canvas.

Text, image and video models side by side. Compare them on the same prompt, pick one per shot, or let the agent choose.

Start flowing.

Write it, picture it, film it. One sentence is enough to begin.