New ByteDance video model, now on Genspark

Seedance 2.5, Online in Genspark

Seedance 2.5 is ByteDance's next-generation video model. Run it in Genspark's video agent — describe a scene below and turn it into a clip of up to 30 seconds, generated in one continuous shot with synchronized audio.

Seedance 2.5 is a model developed by ByteDance. Genspark is an independent AI workspace where you can use Seedance 2.5 alongside other video models.

What's new in Seedance 2.5

Both Seedance generations are selectable in Genspark's video agent. Here is how the 2.5 release compares with Seedance v2, spec for spec, as configured in Genspark.

SpecSeedance 2.5Seedance v2
Max clip length30 seconds, generated in one continuous shot15 seconds
Resolution480p or 720p480p, 720p, or 1080p
Aspect ratiosSix, from 21:9 widescreen to 9:16 verticalSame six ratios
ReferencesUp to 50 files: 30 images, 10 video clips, 10 audio clipsUp to 9 images
AudioSynchronized audio with lip-synced speech, on or off per clipNative audio with lip-sync

Rule of thumb: pick Seedance 2.5 for longer one-take clips and reference-heavy work; stay on Seedance v2 when you need 1080p output.

Prompt recipes to start from

Each row is a complete working prompt. Click one to load it into the box above, edit the details, then send it to the video agent.

How to use Seedance 2.5 in Genspark

1

Describe your scene

Type a prompt above or load a recipe, then send it. You land in Genspark's video agent with your prompt already in the input; sign in first if you're new.

2

Check the model and settings

Your prompt names Seedance 2.5, and the agent picks it up from there. You can also select it yourself in the model list, then set duration, aspect ratio, and audio.

3

Iterate and download

Ask for changes in plain language: retime a shot, swap the setting, regenerate a section. Download the clip when it looks right.

Frequently Asked Questions

What is Seedance 2.5?

Seedance 2.5 is ByteDance's next-generation video model, released in July 2026. Its headline ability is generating a continuous clip of up to 30 seconds in a single pass, with no stitching between shots, plus synchronized audio and reference-image support. It supersedes the earlier Seedance 1.0 and Seedance 1.5 Pro generations.

Is Seedance 2.5 made by Genspark?

No. Seedance is developed by ByteDance. Genspark is an independent AI workspace that gives you a place to run Seedance 2.5 — alongside other video models — in one chat.

How long and what resolution can Seedance 2.5 videos be in Genspark?

Any length from 4 to 30 seconds, at 480p or 720p, in six aspect ratios from 21:9 widescreen to 9:16 vertical. If you need 1080p output, Seedance v2 is available in the same model list.

What's the difference between Seedance 2.5 and Seedance 2.0?

Seedance 2.5 doubles the maximum clip length to 30 seconds, generates it in one continuous shot, and accepts up to 50 reference files, but tops out at 720p. Seedance 2.0, listed as Seedance v2 in Genspark, stays at 15 seconds and 9 reference images while offering 1080p output. Both are selectable in the video agent.

Can I use my own images as references?

Yes. Seedance 2.5 takes up to 50 reference files: 30 images, 10 video clips, and 10 audio clips. It keeps characters, products, voices, and art styles consistent with them, and a single still can become a moving shot — classic image to video.