Seedance 2.5 API
Multimodal video APIHOTSeedance 2.5 is ByteDance's unified multimodal video model for creating or reshaping 4โ30 second clips from text, images, video, and audio references.
Use the live playground to validate prompts, reference assets, native audio, duration, resolution, and the exact request payload before integrating the same workflow into your product.
Lowest cost starts at $0.085/s
480p: 17 credits/s - Video to Video - (input + output) ร 17
480p: 28 credits/s - Text/Image to Video - Output seconds ร 28
720p: 38 credits/s - Video to Video - (input + output) ร 38
720p: 63 credits/s - Text/Image to Video - Output seconds ร 63
USD estimates use $0.005/credit with the best annual plan. Video-reference requests include both input and output duration in billing.
Seedance 2.5 request fields
Review the active Seedance 2.5 schema before sending the same payload from your application.
{
"model": "bytedance/seedance-2-5",
"input": {
"prompt": "A cinematic product reveal with smooth orbit camera movement",
"reference_image_urls": [
"https://example.com/product.png"
],
"reference_video_urls": [],
"reference_audio_urls": [],
"generate_audio": true,
"resolution": "720p",
"aspect_ratio": "16:9",
"duration": 5
}
}Use Seedance 2.5 for generation
The playground, sample request, fields, and pricing on this page follow the selected Seedance 2.5 model.
View detailed API docsOpen the complete documentation for authentication, request fields, task polling, pricing, and response payloads. /models/seedance-2.5.md
promptstringThe text prompt used to generate the video.. ( Max length: 30000 characters)first_frame_urlstringFirst frame image url or asset://{assetId}last_frame_urlstringEnd frame image url or asset://{assetId}reference_image_urlsarrayEnter a list of image URLs or asset://{assetId} (for example: asset://asset-20260404242101-76djj).reference_video_urlsarrayEnter a list of video URLs or asset://{assetId} (for example: asset://asset-20260404242101-76djj) .reference_audio_urlsarrayEnter a list of audio URLs or asset://{assetId} (for example: asset://asset-20260404242101-76djj) .return_last_framebooleanWhether to return the last frame of the video. When draft=true, this parameter cannot be set to true.generate_audiobooleanWhether to generate audio for the video.resolutionenumVideo resolution - 480p for faster generation, 720p for balance. (480p, 720p)aspect_ratioenumVideo aspect ratio configuration. (1:1, 4:3, 3:4, 16:9, 9:16, 21:9, adaptive)durationnumberVideo duration in seconds, from 4 to 30.output_formatenumVideo output format. (mp4, mov)web_searchbooleanEnable online search?nsfw_checkerbooleanDefaults to false. You can set it to false based on your needs. If set to false, our content filtering will be disabled, and all results will be returned directly by the model itself.Create and edit with one Seedance 2.5 API
Direct the subject, movement, camera, timing, and sound with compatible reference assets, then retrieve the finished render through an asynchronous task workflow.
Up to 50 references
Use up to 30 images, 10 video clips, and 10 audio files in a compatible multimodal reference workflow.
4โ30 second output
Choose a precise output duration for short ads, social stories, product sequences, and cinematic concepts.
Reference-led editing
Guide an existing clip toward new motion, performance, camera treatment, pacing, or visual direction.
Native audio
Generate synchronized dialogue, ambience, effects, and music, or guide the result with audio references.
From model test to production request
Validate the exact Seedance 2.5 workflow in three steps before moving it into your application.
1. Build the payload
Add a prompt, choose first/last frames or compatible multimodal references, then set duration, ratio, resolution, and audio.
2. Create the task
Send the model ID and payload to the generation endpoint with your SdanceAI API key.
3. Poll the result
Track the asynchronous task until it succeeds, then save or display the returned MP4 or MOV URL.
What to build with Seedance 2.5 API
Use one model for original generation, reference-driven variation, and video reshaping across repeatable product workflows.
Designed for real application workflows
Account and credit controls
Tasks use your existing SdanceAI account, API key, credit balance, visibility settings, and generation history.
Two reference modes
Use first and last frames together, or use compatible image, video, and audio reference arrays; these modes cannot be combined in one request.
Usage-based pricing
Credits depend on resolution, output duration, and whether video input is present; video-reference billing includes input duration.
Seedance 2.5 API questions
What is Seedance 2.5 API best for?
It is best for multimodal video creation and editing where a product needs repeatable control over subjects, motion, camera language, timing, and native audio.
Which input types can I use?
You can send a prompt, optional first and last frames, or compatible arrays of image, video, and audio references. First/last-frame mode cannot be combined with the multimodal reference arrays.
How many reference assets are supported?
The current Seedance 2.5 schema supports up to 30 reference images, 10 reference videos, and 10 reference audio files in the compatible reference workflow.
What duration and resolution can I request?
Output duration can be set from 4 to 30 seconds. The current model supports 480p and 720p output with common landscape, portrait, square, ultrawide, and adaptive aspect ratios.
Can Seedance 2.5 generate audio?
Yes. Enable native audio for synchronized dialogue, ambience, sound effects, and music. Compatible audio files can also be included as references.
Can I edit an existing video?
Yes. Add a reference video and describe the performance, motion, camera, timing, or visual treatment you want the new render to follow.
How do I receive the finished video?
Create an asynchronous task, then poll the task endpoint using the returned task ID. A successful response includes the generated media URL.
Can I use Seedance 2.5 API commercially?
The page supports commercial workflows. You remain responsible for your rights to prompts and reference media and should review the applicable plan and content-policy terms before production use.
Add Seedance 2.5 to your video workflow
Test a prompt and references in the playground, inspect the generated request, then integrate the same model through the API.