Kling video

Create Kling video tasks through AnyInt's OpenAI-style video route. Use the Transtreams native routes only when you need provider-native payload control.

Kling video generation is an asynchronous task flow. Create a task first, store the returned video ID, then poll the video status route until the task reaches a terminal state.

For new integrations, use the OpenAI-style video route. It accepts public AnyInt model IDs such as kling-3.0 and kling-v3-omni and adapts the request to the underlying Kling provider route.

Base URL

https://api.anyint.ai

Authentication

Authorization: Bearer <ANYINT_API_KEY>

Recommended routes

Route	Purpose
`POST /openai/v1/videos`	Create a Kling video task
`GET /openai/v1/videos/{video_id}`	Query task status
`GET /openai/v1/videos/{video_id}/content`	Download generated video content after completion

Request fields

Field	Type	Required	Meaning	Example
`model`	string	Yes	AnyInt video model ID available to your account. Use `kling-3.0` for Standard/Pro Kling video or `kling-v3-omni` for Omni.	`kling-3.0`
`prompt`	string	Yes	Text instruction describing the generated video.	`A cinematic paper airplane flying through a blue sky.`
`duration`	integer or string	Optional	Target video duration in seconds. Supported values depend on the selected model.	`5`
`aspect_ratio`	string	Optional	Output aspect ratio when supported.	`16:9`
`mode`	string	Optional	Edition selector when supported. Use `std` or `pro`.	`std`
`generate_audio`	boolean	Optional	Whether to request audio when the selected route supports it.	`false`
`sound`	string	Optional	Audio mode for Omni-compatible requests.	`off`
`image`	string	Optional	Public HTTPS image URL used as the first-frame or image-to-video reference.	`https://your-domain.com/source.png`
`input_reference`	string	Optional	Public HTTPS image URL or supported image input used as a reference.	`https://your-domain.com/source.png`
`frame_images`	array	Optional	Structured frame references. Use `role` or `frame_type` to identify the frame role when needed.	`[{"url":"https://...","role":"first_frame"}]`
`input_references`	array	Optional	Additional image, video, or audio references when the selected model supports them.	`[{"url":"https://...","type":"video"}]`

Text-to-video example

curl -X POST "https://api.anyint.ai/openai/v1/videos" \
  -H "Authorization: Bearer $ANYINT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-3.0",
    "prompt": "A cinematic paper airplane flying slowly through a blue sky, stable camera, soft daylight.",
    "duration": "5",
    "aspect_ratio": "16:9",
    "mode": "std",
    "generate_audio": false
  }'

Image-to-video example

Use a publicly reachable HTTPS image URL from your own storage.

curl -X POST "https://api.anyint.ai/openai/v1/videos" \
  -H "Authorization: Bearer $ANYINT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-3.0",
    "prompt": "Animate the subject with a slow cinematic dolly-in while keeping the background consistent.",
    "image": "https://your-domain.com/source-image.png",
    "duration": "5",
    "aspect_ratio": "16:9",
    "mode": "pro",
    "generate_audio": false
  }'

Omni reference example

Use kling-v3-omni when you need the Omni video model.

curl -X POST "https://api.anyint.ai/openai/v1/videos" \
  -H "Authorization: Bearer $ANYINT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-v3-omni",
    "prompt": "Make the character in the reference image turn toward the camera in an animated film style.",
    "input_reference": "https://your-domain.com/person.png",
    "frame_images": [
      {
        "url": "https://your-domain.com/person.png",
        "role": "first_frame"
      }
    ],
    "duration": "5",
    "aspect_ratio": "16:9",
    "sound": "off"
  }'

Create response

Store the returned id. Use it as {video_id} in later status and content requests.

Field	Meaning
`id`	Video task ID to store for polling
`object`	Object type, usually `video`
`status`	Initial task status, such as `processing`
`model`	Model ID used for the task
`error`	Error details when task creation fails

Example response:

{
  "id": "video_...",
  "object": "video",
  "status": "processing",
  "model": "kling-3.0",
  "error": null
}

Query status

curl -X GET "https://api.anyint.ai/openai/v1/videos/$VIDEO_ID" \
  -H "Authorization: Bearer $ANYINT_API_KEY"

Query response fields

Field	Meaning
`id`	Video task ID
`status`	Current task state. Treat `completed` as success and `failed`, `cancelled`, or `canceled` as terminal failure states.
`progress`	Progress value when returned by the provider
`download_url`	Content route for the generated video when the task has completed
`error`	Failure details when the task fails

Example completed response:

{
  "id": "video_...",
  "object": "video",
  "status": "completed",
  "progress": 100,
  "model": "kling-3.0",
  "download_url": "/v1/videos/video_.../content",
  "error": null
}

If download_url is relative, resolve it against the OpenAI-compatible API prefix:

curl -L "https://api.anyint.ai/openai$DOWNLOAD_URL" \
  -H "Authorization: Bearer $ANYINT_API_KEY" \
  --output kling-video.mp4

Transtreams native routes

The provider-native Transtreams routes remain available when you need the native request shape:

Route	Purpose
`POST /transtreams/kling/v1/videos/text2video`	Create a native text-to-video task
`POST /transtreams/kling/v1/videos/image2video`	Create a native image-to-video task
`GET /transtreams/kling/v1/videos/text2video/{task_id}`	Query a native text-to-video task
`GET /transtreams/kling/v1/videos/image2video/{task_id}`	Query a native image-to-video task
`POST /transtreams/kling/v1/videos/omni-video`	Create a native Kling V3 Omni task
`GET /transtreams/kling/v1/videos/omni-video/{task_id}`	Query a native Kling V3 Omni task

Use native routes only when your integration already depends on Transtreams-native fields such as model_name, image_list, or video_list. New integrations should prefer /openai/v1/videos.

For native Standard/Pro requests, use model_name: "kling-v3" with mode: "std" or "pro". For native Omni requests, use model_name: "kling-v3-omni" and the /videos/omni-video route.

Common mistakes

Sending Kling video requests to Chat Completions or Responses routes.
Using /transtreams/kling/... for a new integration when /openai/v1/videos is sufficient.
Sending kling-v3-std as native model_name instead of model_name: "kling-v3" plus mode: "std".
Sending kling-v3-omni to a native Standard/Pro route instead of /transtreams/kling/v1/videos/omni-video.
Treating a processing response as a completed video. Always poll until a terminal status.
Fetching a relative download_url without the /openai prefix or without the Authorization header.

On this page