Seedance 2.5 Flash Video Generation API
Generate asynchronous videos from text, images, video or audio references. Use the exact public model ID `seedance-2.5-flash`; it has its own prices and limits.
Seedance 2.5 Flash Video Generation API
Generate asynchronous videos from text, images, video or audio references. Use the exact public model ID seedance-2.5-flash; it has its own prices and limits.
Capabilities and limits
Output supports 480p / 720p and integer durations 4–30 seconds, default 5. duration: -1 enables automatic duration. 1080p, 4K and Draft are not supported.
Up to 10 images, 5 videos and 5 audio clips per request. Each reference video/audio is 2–30 seconds; combined video duration and combined audio duration must each stay within 30 seconds. An image must not exceed 30 MB; a video 200 MB; an audio clip 15 MB. Use authorized assets and reachable URLs.
Parameters
| Parameter | Type | Value / limit |
|---|---|---|
model |
string | seedance-2.5-flash |
prompt |
string | Text |
resolution |
string | 480p / 720p; default 720p |
duration |
integer | 4–30, -1; default 5 |
seconds |
integer / string | Alias: duration |
aspect_ratio |
string | 16:9, 4:3, 1:1, 3:4, 9:16, 21:9, adaptive; default adaptive |
ratio |
string | Alias: aspect_ratio |
size |
string | Alias: resolution / pixel size; no conflicting resolution or ratio |
generate_audio |
boolean | default true |
audio |
boolean | Alias: generate_audio |
watermark |
boolean | default false |
return_last_frame |
boolean | default false |
seed |
integer | 0 is preserved |
output_format |
string | mp4 / mov; default mp4 |
omni_reference_task_type |
string | auto, reference, edit, extend |
image_urls |
string[] | 10 |
image_with_roles |
object[] | url, role: first_frame, last_frame, reference_image; images total ≤10 |
video_urls |
string[] | 5 |
audio_urls |
string[] | 5 |
content |
object[] | Native text, image_url, video_url, audio_url references |
metadata |
object | Optional wrapper for generation fields |
draft |
boolean | true: unsupported |
draft_task_id |
string | unsupported |
Keep adaptive for first/last-frame, edit and extend tasks; output follows the reference proportions. Editing requires a 4–30 second source video and duration: -1. Reference audio may be used on its own. MOV and other advanced options depend on the selected generation route.
Submit and download
Save the public task ID returned by submission, poll until completion, then download the video. HTTP 200 means accepted, not completed. On failure, read the error before starting another task. Reuse an Idempotency-Key only when retrying the identical request.
curl "https://apimaster.ai/v1/v1/videos" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: flash-text-unique-001" \
-d '{"model":"seedance-2.5-flash","prompt":"A red paper boat floating on calm water at sunrise","duration":4,"resolution":"720p","aspect_ratio":"16:9","generate_audio":true,"watermark":false}'
curl "https://apimaster.ai/v1/v1/videos/TASK_ID" -H "Authorization: Bearer YOUR_API_KEY"
curl -L "https://apimaster.ai/v1/v1/videos/TASK_ID/content" -H "Authorization: Bearer YOUR_API_KEY" -o video.mp4
curl "https://apimaster.ai/v1/v1/videos/TASK_ID/billing" -H "Authorization: Bearer YOUR_API_KEY"
References, editing and extension
{
"model": "seedance-2.5-flash",
"prompt": "The paper boat passes under a wooden bridge",
"duration": 4,
"resolution": "480p",
"aspect_ratio": "adaptive",
"generate_audio": false,
"image_with_roles": [
{
"url": "https://example.com/authorized-first-frame.png",
"role": "first_frame"
}
]
}
{
"model": "seedance-2.5-flash",
"prompt": "Change the lighting to warm sunset",
"resolution": "720p",
"duration": -1,
"aspect_ratio": "adaptive",
"omni_reference_task_type": "edit",
"video_urls": [
"https://example.com/authorized-source.mp4"
]
}
omni_reference_task_type: "extend" + duration: 4 + aspect_ratio: "adaptive".
Billing
Base prices below use the international official token rate as the reference. Actual customer charges also apply the current channel, group and account settings; check the marketplace card and the task billing receipt. Image-only requests bill output seconds. Video references bill measured input seconds plus output seconds. Official token billing may also have minimum usage floors; actual provider charges and platform second-based estimates can differ.
| Resolution | USD / second | USD / second, video input |
|---|---|---|
| 480p | 0.0672525 | 0.04131225 |
| 720p | 0.1512 | 0.09288 |
Troubleshooting
Use the public model ID, check resolution and reference limits, and keep JSON booleans such as false explicit. A missing fingerprint baseline is not evidence that generation failed. Download results promptly; do not depend on temporary video URLs for long-term storage.
