AI video modelSeedance 2.5 AI Video Generator
ByteDance’s Seedance model for prompt-faithful motion, with reference images for consistent people.
Opens the generator with this model and the example prompt selected.
Specs
- Length
- 4–30s
- Resolution
- 480P · 720P · 1080P
- Inputs
- Text to video · Image to video · First & last frame · Reference to video
- Max input images
- 2
- Price
- From 540 credits per video
About Seedance 2.5
Seedance 2.5 is the newest Seedance video model in our catalog. It follows long prompts closely, keeps camera moves smooth, and can take up to two reference images so the same people appear in the clip.
We use it for our two-person effects (kiss, hug, couple and duet trends) because reference-to-video keeps both faces apart instead of blending them.
Example prompt
A slow dolly-in on two friends laughing at a rooftop café at golden hour, warm backlight, soft lens flare, natural handheld feel.
Best for
- Two-person scenes that need both identities kept
- Longer clips up to 30 seconds
- Cinematic camera moves described in the prompt
Frequently asked questions
Other video models
The proven Seedance model for clean motion and first-to-last-frame transitions.
4–15s · 480P / 720P / 1080P · From 370 credits per video
Expressive, natural motion from one photo.
3 / 5 / 10 / 15s · 720P / 1080P · From 200 credits per video
An affordable model for short, well-composed clips.
2 / 3 / 4 / 5 / 10 / 15s · 480P / 720P / 1080P · From 80 credits per video
Google’s cinematic video model, with a Fast tier for drafts.
4 / 6 / 8s · 720P / 1080P · From 760 credits per video
Sharp 2K output and clips up to 15 seconds.
5–15s · 768P / 2K · From 280 credits per video