curl --request POST \
--url https://gengen.farm/api/gengen/v1/contents/generations/tasks \
--header 'Authorization: Bearer gengen_live_xxxxxxxxxxxxxxxx' \
--header 'Content-Type: application/json' \
--data '{
"model": "dreamina-seedance-2-0-260128",
"mode": "text_to_video",
"prompt": "A kitten yawns at the camera in warm morning light.",
"contentPreFilter": true,
"controls": {
"duration": 5,
"resolution": "720p",
"ratio": "16:9",
"generateAudio": true,
"watermark": false
}
}'{
"id": "seedance20:cgt-20260419-abcdef1234567890",
"model": "dreamina-seedance-2-0-260128",
"status": "pending",
"outputs": {
"videos": [],
"images": []
}
}{
"error_code": "gengen.model_required",
"error_params": {
"field": "model"
},
"error": "A model is required"
}{
"error_code": "gengen.auth_required",
"error": "Missing Authorization bearer token"
}{
"error_code": "seedance.insufficient_balance",
"error": "Insufficient available balance"
}{
"error_code": "gengen.api_key_revoked",
"error": "GENGEN API key has been revoked"
}{
"error_code": "<string>",
"error_params": {
"failedAsset": {
"url": "<string>",
"field": "assets.referenceImages",
"index": 1,
"mediaType": "Image",
"stage": "validation"
}
},
"error": "<string>"
}Create a video generation task
Submit an asynchronous BytePlus Seedance video generation task. This
reference intentionally documents only the BytePlus models and parameters,
even though other providers use the same public GENGEN endpoint.
Poll the returned ID for successful Seedance token counts in usage.
HTTP(S) reference images, videos and audio are automatically registered in
your API asset group before generation. Existing asset:// references are
reused subject to access checks. Requests containing URLs return a stable
task ID with status: pending and stage: preparingAssets; poll that ID.
Text-only and all-asset-URI requests keep their existing submission flow.
Automatic registration requires HTTP(S) media; inline/base64 inputs are not
supported on this path. Each new POST is a separate generation request.
If preparation fails, task retrieval returns status: failed and
error.failedAsset with the original URL, canonical field, zero-based index,
media type and failing stage. No video generation charge is applied.
Seedance 2.5 Flash (dreamina-seedance-2-5-flash-261115) supports 4–30
second output at 480p/720p, up to 10 images, 5 videos (including the source),
and 5 audio clips. Audio-only references are supported. Flash does not
support Draft, 1080p, 4K, or flex. Its duration, ratio, task-type, and MP4/MOV
controls follow the 2.5 series. Video/audio clips are 2–30 seconds each
(4–30 for video editing), at most 30 seconds total per media type.
For Seedance 2.5, each reference video or audio clip must be 2–30 seconds long. You can provide up to 10 reference videos and 10 reference audio clips. The combined duration of all reference videos and the combined duration of all reference audio clips must each stay within 30 seconds.
For Seedance 2.5 omni-reference workflows, an explicitly supplied
multimodal_reference, video_modify, or video_extend mode selects
reference generation, editing, or extension. If mode is omitted,
the task type is determined automatically. Edit and
extend requests require prompts with clear matching intent.
curl --request POST \
--url https://gengen.farm/api/gengen/v1/contents/generations/tasks \
--header 'Authorization: Bearer gengen_live_xxxxxxxxxxxxxxxx' \
--header 'Content-Type: application/json' \
--data '{
"model": "dreamina-seedance-2-0-260128",
"mode": "text_to_video",
"prompt": "A kitten yawns at the camera in warm morning light.",
"contentPreFilter": true,
"controls": {
"duration": 5,
"resolution": "720p",
"ratio": "16:9",
"generateAudio": true,
"watermark": false
}
}'{
"id": "seedance20:cgt-20260419-abcdef1234567890",
"model": "dreamina-seedance-2-0-260128",
"status": "pending",
"outputs": {
"videos": [],
"images": []
}
}{
"error_code": "gengen.model_required",
"error_params": {
"field": "model"
},
"error": "A model is required"
}{
"error_code": "gengen.auth_required",
"error": "Missing Authorization bearer token"
}{
"error_code": "seedance.insufficient_balance",
"error": "Insufficient available balance"
}{
"error_code": "gengen.api_key_revoked",
"error": "GENGEN API key has been revoked"
}{
"error_code": "<string>",
"error_params": {
"failedAsset": {
"url": "<string>",
"field": "assets.referenceImages",
"index": 1,
"mediaType": "Image",
"stage": "validation"
}
},
"error": "<string>"
}Body
Normalized GENGEN request for BytePlus Seedance video models. Other models share this endpoint but have separate reference pages and request parameter sets. Seedance 2.5 accepts up to 10 reference videos and 10 reference audio clips. Each clip must be 2–30 seconds long, and each media type has a separate 30-second combined-duration limit.
Seedance 2.5 Flash supports up to 10 images, 5 videos including the source, and 5 audio clips, with the same duration limits. Flash supports only 480p/720p and does not support Draft workflows.
For Seedance 2.5 (including Flash), explicitly supplied omni-reference modes are converted to
BytePlus task-type hints. Omitted modes retain BytePlus's default auto
classification behavior.
Supported BytePlus Seedance model IDs:
dreamina-seedance-2-5-260628— Seedance 2.5dreamina-seedance-2-5-flash-261115— Seedance 2.5 Flashdreamina-seedance-2-0-260128— Seedance 2.0dreamina-seedance-2-0-fast-260128— Seedance 2.0 Fastdreamina-seedance-2-0-mini-260615— Seedance 2.0 Miniseedance-1-5-pro-251215— Seedance 1.5 Pro
dreamina-seedance-2-5-260628, dreamina-seedance-2-5-flash-261115, dreamina-seedance-2-0-260128, dreamina-seedance-2-0-fast-260128, dreamina-seedance-2-0-mini-260615, seedance-1-5-pro-251215 "dreamina-seedance-2-0-260128"
Generation mode. If omitted, GENGEN attempts to infer the mode from
the supplied assets for request construction, but does not send an
explicit Seedance 2.5 task-type hint; BytePlus therefore keeps its
default auto classification behavior. Seedance 2.5, 2.0, 2.0 Fast,
and 2.0 Mini support all six values. Seedance 1.5 Pro supports only
text_to_video, image_first_frame, and image_first_last_frame.
Supported values:
text_to_video— generate from a text promptimage_first_frame— animate a first-frame imageimage_first_last_frame— interpolate between first and last framesmultimodal_reference— guide generation with image, video, or audio references; maps to Seedance 2.5 task typereferencevideo_modify— edit an existing source video; maps to Seedance 2.5 task typeeditvideo_extend— extend an existing source video; maps to Seedance 2.5 task typeextend
Explicit Seedance 2.5 omni-reference modes enable submission-time
validation of their task-specific constraints. BytePlus still checks
prompt intent during processing. A conflicting prompt can fail
asynchronously with InvalidParameter.TaskTypeMismatch. Automatic
classification can return InvalidParameter.TaskTypeConstraint
asynchronously when the inferred task conflicts with supplied controls.
text_to_video, image_first_frame, image_first_last_frame, multimodal_reference, video_modify, video_extend Creative text instruction. Required for text_to_video, video_modify,
and video_extend. A video-editing prompt must clearly describe the
intended edit, such as adding, removing, replacing, or changing content.
A video-extension prompt must clearly ask to extend or continue the
source video. Recommended for every other BytePlus Seedance mode.
Refer to audio clips as Audio 1, Audio 2, etc., following their
one-based order in assets.referenceAudios, independently of images
and videos.
Response
The task was accepted or completed synchronously by the selected provider.
Standard task fields are listed first. Additional provider-native and legacy aliases can be appended for compatibility.
Optional informational notice for a Seedance task that is still processing more than 420 seconds after upstream creation. This notice is omitted otherwise; existing upstream passthrough messages are unchanged.
"Video generation is taking longer than usual. Seedance may be experiencing high demand. Your video is still generating."