Video Generation API

Start AI video generations, poll their status, and list, download and delete videos with the Kensa API.

All paths are relative to https://www.kensa.cc/api/v1 and need an API key (see API Overview), except List models.

Generate a video

POST /video/generate

curl -X POST https://www.kensa.cc/api/v1/video/generate \
  -H "Authorization: Bearer kens_your_key" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "veo-3.1",
    "prompt": "A golden retriever running through a sunlit meadow, cinematic slow motion",
    "aspectRatio": "16:9"
  }'

Request body

Field	Type	Required	Description
`model`	string	Yes	Model ID: `seedance-2.5`, `seedance-2.0`, `veo-3.1`, `seedance-1.5-pro`, `grok-image-video` or `minimax-h3`
`prompt`	string	Yes	What the video should show, 1 to 5,000 characters
`duration`	number	No	Length in seconds; one of the model's durations below. Defaults to the shortest.
`aspectRatio`	string	No	One of the model's aspect ratios below. Defaults to `16:9`.
`quality`	string	No	Resolution: `480p`, `720p` or `1080p`, as the model supports
`imageUrl`	string	No	One `https` image URL (JPG, PNG or WebP, up to 20 MB) to use as the first frame
`imageUrls`	string[]	No	Several `https` image URLs, up to 9 per request. With `seedance-2.5`, `seedance-2.0`, `veo-3.1` and `seedance-1.5-pro`, two images are the first and last frame. Use either `imageUrl` or `imageUrls`.
`mode`	string	No	`reference-to-video` to use `imageUrls` as reference images of the subject instead of frames (`seedance-2.5`, `seedance-2.0` and `veo-3.1` only)
`generateAudio`	boolean	No	Seedance models generate audio unless you send `false`

Options per model

`model`	`duration`	`aspectRatio`	`quality`
`seedance-2.5`	4, 5, 6, 8, 10, 12, 15, 20, 30	16:9, 9:16, 1:1, 4:3, 3:4, 21:9 (image-to-video follows the image)	`480p` (default), `720p`, `1080p` (paid plans)
`seedance-2.0`	4, 5, 6, 8, 10, 12, 15	16:9, 9:16, 1:1, 4:3, 3:4, 21:9	`480p` (default), `720p`
`veo-3.1`	Always 8	16:9, 9:16, 1:1	Not needed
`seedance-1.5-pro`	4, 5, 6, 8, 10, 12	16:9, 9:16, 1:1, 4:3, 3:4, 21:9	`480p`, `720p` (default), `1080p`
`grok-image-video`	6, 10	16:9, 9:16, 1:1, 4:3, 3:4	`480p` (default), `720p`
`minimax-h3`	Any whole number from 5 to 15	16:9, 9:16, 1:1	`480p`, `768p` (default), `2k` (paid plans)

kling-3 is under maintenance; requests for it return 503 with details.reason: "model_maintenance". Prices are on AI Models.

Before a generation starts, the prompt and every image are checked against the usage policy. A request that fails the check, or whose image can't be read, returns 400 and costs nothing.

Response

{
  "success": true,
  "data": {
    "videoUuid": "vid_V1StGXR8_Z5jdHi6B-myT",
    "taskId": "task_8f2c1e",
    "provider": "evolink",
    "status": "GENERATING",
    "creditsUsed": 13
  }
}

creditsUsed is the exact price. The credits are reserved when the generation starts and returned if it fails. Use videoUuid for the other endpoints.

Image examples

First and last frame:

{
  "model": "seedance-2.0",
  "prompt": "The flower bud slowly opens into a full bloom, soft morning light",
  "duration": 5,
  "aspectRatio": "9:16",
  "imageUrls": ["https://example.com/bud.jpg", "https://example.com/bloom.jpg"]
}

Reference images:

{
  "model": "veo-3.1",
  "mode": "reference-to-video",
  "prompt": "The character from the reference images walks through a rainy night market",
  "imageUrls": ["https://example.com/front.png", "https://example.com/side.png"]
}

Check the status

GET /video/{uuid}/status

{
  "success": true,
  "data": {
    "status": "COMPLETED",
    "videoUrl": "https://..."
  }
}

A failed generation returns "status": "FAILED" and an error message. Poll every 10 to 15 seconds; most clips are ready in a few minutes. A generation that hasn't finished after 2 hours is marked FAILED and its credits are returned.

Status	Meaning
`PENDING`	Created, waiting for the model
`GENERATING`	The model is generating the video
`UPLOADING`	The video is being saved
`COMPLETED`	Ready; `videoUrl` is set
`FAILED`	Failed; the reserved credits are returned

Get a video

GET /video/{uuid}

Returns the video, including uuid, status, prompt, model, videoUrl, thumbnailUrl, duration, aspectRatio, creditsUsed, errorMessage, createdAt and completedAt. Returns 404 if the video doesn't exist or was deleted.

Download a video

GET /video/{uuid}/download

Redirects (302) to a temporary download link for the MP4 file. Returns 409 while the file isn't ready.

curl -L -o video.mp4 https://www.kensa.cc/api/v1/video/vid_V1StGXR8_Z5jdHi6B-myT/download \
  -H "Authorization: Bearer kens_your_key"

GET /video/list?limit=20

Parameter	Default	Description
`limit`	20	Videos per page
`cursor`		The `nextCursor` value from the previous page
`status`		Only videos with this status, for example `COMPLETED`

{
  "success": true,
  "data": {
    "videos": [{ "uuid": "vid_V1StGXR8_Z5jdHi6B-myT", "status": "COMPLETED", "model": "veo-3.1" }],
    "nextCursor": "vid_Uakgb_J5m9g-0JDMbcJqL"
  }
}

Videos are sorted newest first. nextCursor is missing on the last page.

Delete a video

DELETE /video/{uuid}

Removes the video from your list. Returns { "success": true, "data": { "deleted": true } }.

List models

GET /config/models

No API key needed. Returns each model with its id, name, supportImageToVideo, durations, aspectRatios, qualities and creditCost. kling-3 is listed while it is under maintenance, but generating with it returns 503.