Text to Video
16:9An octopus on the sandy seafloor wrapping its arms around a soccer ball, clear blue water, realistic underwater physics
Seedance 2.0 Fast
ID:
bytedance/seedance-2-fastSeedance 2.0 Fast video generation API by ByteDance — speed-optimized variant with audio generation, adaptive aspect ratio, and clips up to 15 seconds.
Text to videoImage to videoAudio generationFast
Input
OutputView all
Output will appear here
Pricing
480p
$0.10/s
720p
$0.21/s
Examples
Overview
Seedance 2.0 Fast is the speed-optimized variant of ByteDance's second-generation video model, released in February 2026. It trades maximum resolution (capped at 720p instead of 1080p) for faster generation times, making it suited for iteration-heavy workflows. The model retains the core 2.0 capabilities — native audio generation, 4-to-15-second variable duration, and a 20000-character prompt limit — and adds an adaptive aspect ratio option that lets the model choose framing based on the prompt content.
Use cases
Rapid iteration and batch generation when turnaround matters more than peak resolution. Quick prototyping of video concepts before committing to the full Seedance 2.0 model. Social media drafts and internal review cuts. Exploratory workflows using adaptive aspect ratio to let the model pick the best framing.
Inputs
All parameters are passed in the input object of the run request.
| Parameter | Required | Description |
|---|---|---|
| prompt | Yes | Text description (3–20000 chars) |
| aspect_ratio | No | Default 16:9. Options: 1:1, 3:4, 4:3, 9:16, 16:9, 21:9, adaptive |
| resolution | No | Default 720p. Options: 480p, 720p |
| duration | No | Video length in seconds (4–15). Default 5 |
| generate_audio | No | Generate audio track. Default false |
| image_urls | No | First frame image (max 1, max 10 MB) for image-to-video |
Prompt tips
Try adaptive for open-ended prompts
When you don't have a specific aspect ratio in mind, set aspect_ratio to adaptive. The model analyzes your prompt and picks the framing that fits the scene — landscape for vistas, portrait for single subjects.
Use short durations for single actions
The 4-to-15-second range is flexible. For a single motion (a pour, a jump, a turn), 4–6 seconds keeps the output tight. Save longer durations for sequences with multiple beats.
Limitations
- Maximum resolution is 720p — no 1080p option (use standard Seedance 2.0 for that)
- Audio generation increases credits cost
- Prompt minimum is 3 characters
- Only one first-frame image allowed for image-to-video (max 10 MB)
- Lower visual detail than the standard Seedance 2.0 at the same 720p setting
FAQ
How much faster is it than standard Seedance 2.0?
Generation times vary by load and parameters, but Fast is optimized for lower latency at the cost of the 1080p resolution tier. For the same 720p output, expect noticeably shorter wait times.
Does it support audio like the standard 2.0?
Yes. Set generate_audio to true for synchronized dialogue, sound effects, and ambient audio. This works identically to the standard Seedance 2.0 model.
What does adaptive aspect ratio do?
Instead of locking to a fixed ratio, the model infers the best framing from your prompt. A landscape scene might get 16:9; a standing portrait might get 9:16. It's useful when you want the model to decide.