AI Video APIMore context in one request
The larger image, video, and audio limits let applications express complex creative direction without splitting the same scene across multiple generation jobs.
Next-generation multimodal video API
Seedance 2.5 API
Create up to 30 seconds of video from text, frames, images, video, or audio references.
Seedance 2.5 expands multimodal video generation with longer 4–30 second output, up to 30 image references, 10 video references for fixed-duration requests, 10 audio references, and audio-only input. With duration=-1, video count and combinations are validated by the upstream. Use one asynchronous API for text-to-video, reference-guided creation, first/last-frame control, synchronized audio, MP4 or MOV output, callbacks, and task polling.
4–30s / -1
Output duration
480p / 720p / 1080p
Resolution
30 images + 10 audio + upstream-validated video
Reference capacity
ByteDance
doubao-seedance-2.5async task
4–30s / -1
Output duration
480p / 720p / 1080p
Resolution
30 images + 10 audio + upstream-validated video
Reference capacity
Model capabilities
Built for production API workflows, not one-off demos.
Use Seedance 2.5 from the same platform surface as the rest of your video and image stack: API keys, credits, logs, webhooks, and docs stay consistent across providers.
High-capacity multimodal references
Fixed-duration requests can combine up to 30 images, 10 videos, and 10 audio clips. With duration=-1, video count and combinations are validated by the upstream.
Audio-only video generation
Use a music, voice, or sound reference without an image or video input. Seedance 2.5 can build the visual sequence around the supplied audio intent.
First and last frame control
Provide one image as a first frame or two images as first and last frames. The dedicated frame mode uses adaptive ratio and preserves the intended transition boundaries.
Longer clips with synchronized audio
Generate 4–30 second clips with synchronized audio enabled by default, giving campaigns and narrative sequences more room than the previous 15-second workflow.
MP4, MOV, and reusable last frames
Choose MP4 or MOV output and optionally request the generated last frame, making it easier to continue a scene or integrate the result into an editing pipeline.
Production-ready asynchronous API
Submit through the shared video endpoint, receive a task ID immediately, and obtain completion or failure through polling or a callback URL using the platform's standard task response.
API workflow
Submit tasks, track progress, and return generated assets.
01
Choose a generation mode
Use omni_reference for text or any supported media combination, including audio-only input. Use first_and_last_frames with one or two images for frame-controlled generation.
02
Configure the output
Select a 4–30 second duration, or use duration=-1 with or without video input and let the upstream determine output length. Configure 480p, 720p, or 1080p, aspect ratio, synchronized audio, watermark behavior, and MP4 or MOV output.
03
Submit the asynchronous task
POST to the unified video generation endpoint with model doubao-seedance-2.5. Reference-video duration is detected before charging so the correct pricing tier can be applied.
04
Poll or receive a callback
Track GET /v1/tasks/{taskId} or provide callback_url. Completed tasks return hosted output URLs and an optional last-frame URL in the standard response shape.
POST /v1/videos/generations
curl -X POST https://api.aivideoapi.ai/v1/videos/generations \
-H "Authorization: Bearer sk-your-api-key" \
-H "Content-Type: application/json" \
-d '{
"model": "doubao-seedance-2.5",
"callback_url": "https://your-server.com/webhooks/video",
"input": {
"prompt": "Keep the character design, follow the reference camera rhythm, and use the audio for ambience",
"image_urls": ["https://example.com/character.png"],
"video_urls": ["https://example.com/camera-reference.mov"],
"audio_urls": ["https://example.com/ambience.wav"],
"duration": 10,
"resolution": "720p",
"aspect_ratio": "adaptive",
"generate_audio": true,
"output_format": "mp4"
}
}'What teams build
30-second campaign videos
Audio-driven visual sequences
First-to-last-frame transitions
Character and product consistency
Reference motion and camera transfer
Music and ambience-led storytelling
Storyboard and concept development
Scene continuation workflows
Why choose Seedance 2.5
The Seedance 2.5 API benefits that matter to developers.
Audio can lead the workflow
Unlike workflows that require a visual reference alongside audio, Seedance 2.5 accepts audio-only input and can use sound as the starting point for the generated video.
Transparent reference-video billing
The platform measures reference-video duration before task creation. When video is supplied, both detected input seconds and requested output seconds use the lower video-reference rate.
Frequently asked questions
Answers about the Seedance 2.5 API.
What is the Seedance 2.5 API?
The Seedance 2.5 API is AI Video API's asynchronous integration for ByteDance's multimodal video model. Use model doubao-seedance-2.5 with text, image, video, or audio inputs to create 4–30 second videos at 480p, 720p, or 1080p.
Which Seedance 2.5 generation modes are supported?
Use omni_reference for text-to-video or any supported combination of image, video, and audio references. Audio may be used alone. Use first_and_last_frames with one image for a first frame or two images for first and last frames; that mode cannot include video or audio references.
How many references can Seedance 2.5 accept?
One request supports up to 30 images, 10 videos with no more than 30 seconds total duration, and 10 audio clips with no more than 30 seconds total duration. First-and-last-frame mode accepts one or two images.
Can Seedance 2.5 generate video from audio only?
Yes. In omni_reference mode, audio_urls can be the only media input and the prompt may be omitted. The model can use music, voice, or sound as the creative basis for the visual sequence.
How much does Seedance 2.5 cost?
Without video input, Seedance 2.5 costs 25.86 credits per output second at 480p, 58.17 at 720p, and 103.68 at 1080p. With video input, it costs 17.24 credits per billable second at 480p, 38.78 at 720p, and 69.12 at 1080p. For duration=-1, final billing uses actual output seconds without video, or input-video seconds plus actual output seconds with video.
Which output formats does Seedance 2.5 support?
Seedance 2.5 supports MP4 and MOV output. You can also set return_last_frame to true to receive the generated video's final frame for chaining or editing workflows.
What happens after a Seedance 2.5 request is submitted?
The API validates the request, measures reference-video duration when necessary, pre-charges the calculated credits, and returns a task ID. Poll the task endpoint or use callback_url. Confirmed failed or expired tasks receive a full refund; completed tasks retain the final charge.
Pricing and usage
Clear model options with shared credits.
480p without video input
25.86 credits/s
Output duration only
480p with video input
17.24 credits/s
Input-video plus output duration
720p without video input
58.17 credits/s
Default resolution; output duration only
720p with video input
38.78 credits/s
Input-video plus output duration
1080p without video input
103.68 credits/s
Output duration only
1080p with video input
69.12 credits/s
Input-video plus output duration
Start building with Seedance 2.5 in AI Video API.
Create one API key, use one credit balance, and switch between video and image models without provider-specific plumbing.