Seedance 2.5 Flash
The fast tier of Seedance 2.5, with 4–30 second videos, synchronized audio, and multimodal references.
| Model ID | Aliases | Endpoint |
|---|---|---|
bytedance/seedance-2.5-flash |
seedance-2.5-flash, dreamina-seedance-2-5-flash, doubao-seedance-2-5-flash |
Video generation |
Input modes
| Mode | References | Reference roles |
|---|---|---|
text Text only |
— | — |
image Image input |
1 to 2 | first_frame 1last_frame 0 to 1 |
reference Reference media |
1 to 20 | reference_audio 0 to 5reference_image 0 to 10reference_video 0 to 5 |
Parameters
| Parameter | Description | Values | Default | Required |
|---|---|---|---|---|
duration |
Duration: Output video duration in seconds. Only in text, image. |
4 to 30 |
5 |
No |
duration |
Duration: Output video duration in seconds. With a reference video, “auto” follows the input video for edits; aspect_ratio must also be auto. task_type edit implies auto. Only in reference. |
4 to 30, auto |
5 |
No |
resolution |
Resolution: Output resolution. | 480p, 720p |
720p |
No |
aspect_ratio |
Aspect ratio: Output aspect ratio. Only in text, reference. |
auto, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16 |
auto |
No |
aspect_ratio |
Aspect ratio: Output aspect ratio. Only in image. |
auto |
auto |
No |
return_last_frame |
Return last frame: Return the final video frame as an additional image when supported by the selected supply. | false, true |
false |
No |
generate_audio |
Generate audio: Generate native audio with the video. | false, true |
true |
No |
n |
Outputs per request: Number of outputs from one API request (one Job), not the application batch size. For multiple Jobs, submit separate requests; n=1 does not prevent application-managed batching. | 1 |
1 |
No |
task_type |
Task type: How the references are used. auto (default) lets the model decide from the references and prompt. reference generates a new video guided by them. edit changes the reference video (4 to 30 seconds); its duration and aspect ratio follow that video. extend continues the reference video; aspect_ratio follows it and duration is the length of the new part. The model still checks the prompt, so a prompt that reads as an edit (for example “replace” or “copy exactly”) can fail as reference; edit and extend run on fewer channels. Only in reference. |
auto, reference, edit, extend |
auto |
No |
Examples
text input:
{
"model": "bytedance/seedance-2.5-flash",
"input": {
"type": "text",
"prompt": "A paper boat drifting down a rainy city street, slow tracking shot"
},
"parameters": {
"duration": 5,
"resolution": "720p",
"aspect_ratio": "auto",
"return_last_frame": false,
"generate_audio": true,
"n": 1
}
}
image input:
{
"model": "bytedance/seedance-2.5-flash",
"input": {
"type": "image",
"prompt": "A paper boat drifting down a rainy city street, slow tracking shot",
"references": [
{
"role": "first_frame",
"file_id": "file_..."
}
]
},
"parameters": {
"duration": 5,
"resolution": "720p",
"aspect_ratio": "auto",
"return_last_frame": false,
"generate_audio": true,
"n": 1
}
}
reference input:
{
"model": "bytedance/seedance-2.5-flash",
"input": {
"type": "reference",
"prompt": "A paper boat drifting down a rainy city street, slow tracking shot",
"references": [
{
"role": "reference_image",
"file_id": "file_..."
}
]
},
"parameters": {
"duration": 5,
"resolution": "720p",
"aspect_ratio": "auto",
"return_last_frame": false,
"generate_audio": true,
"n": 1,
"task_type": "auto"
}
}
The authoritative contract is GET /v1/models/bytedance/seedance-2.5-flash?view=contract. A valid parameter does not mean every channel can run it, so quote the complete request. Prices and channels are listed in Model Market.