Last updated: 2026-06-24

AIGC Kling VOD Guide

aigc-video-kling-* models are Kling video models exposed through Crazyrouter’s Tencent VOD route. They do not use the native Kling /kling/v1/videos/* contract. Use the OpenAI-style async video route instead:

POST /v1/video/generations
GET /v1/video/generations/{task_id}

To use the VOD route, submit an aigc-video-kling-* model name. Do not submit kling-v2-5-turbo, kling-v2-6, or kling-v3 here; those names may route to the native Kling provider.

Available Models

Model	Upstream version	Typical capabilities
`aigc-video-kling-1.6`	Kling 1.6	text-to-video, image-to-video, first-last frames, references
`aigc-video-kling-2.0`	Kling 2.0	text-to-video, image-to-video, first-last frames, references
`aigc-video-kling-2.1`	Kling 2.1	text-to-video, image-to-video, first-last frames, references
`aigc-video-kling-2.5-turbo`	Kling 2.5 Turbo	text-to-video, image-to-video, first-last frames, references
`aigc-video-kling-2.6`	Kling 2.6	text-to-video, image-to-video, first-last frames, references, video input
`aigc-video-kling-2.6-motion-control`	Kling 2.6 Motion Control	motion control
`aigc-video-kling-3.0`	Kling 3.0	text-to-video, image-to-video, first-last frames, references
`aigc-video-kling-3.0-turbo`	Kling 3.0 Turbo	text-to-video, image-to-video, references
`aigc-video-kling-3.0-motion-control`	Kling 3.0 Motion Control	motion control
`aigc-video-kling-o1`	Kling O1	text-to-video, image-to-video, references, video input
`aigc-video-kling-avatar`	Kling Avatar	avatar / digital human
`aigc-video-kling-identifyface`	Kling Identifyface	lip sync

GV 3.1, 3.1-fast, and 3.1-lite in Tencent VOD are Google Veo models, not Kling 3.1. Use the separate aigc-video-gv-* models.

Create a Task

POST https://api.crazyrouter.com/v1/video/generations
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

Common Parameters

Parameter	Type	Required	Description
`model`	string	Yes	One of the `aigc-video-kling-*` model names above
`prompt`	string	Conditionally	Required for text-to-video; some multi-shot or image workflows can place prompt data in metadata
`seconds` / `duration`	string / number	No	Duration, usually `5` or `10`; defaults to 5 seconds when omitted
`size`	string	No	For example `1280x720` or `720x1280`; used to infer aspect ratio
`image`	string	No	First-frame URL for single image-to-video
`images`	array[string]	No	First item is the first frame; second item can be a last frame or reference
`metadata`	object	No	VOD Kling extension fields

Common `metadata` Fields

Field	Description
`resolution`	Output tier such as `720P`, `1080P`, `2K`, or `4K`
`aspect_ratio`	Output ratio such as `16:9`, `9:16`, or `1:1`
`sound`	Audio generation toggle: `true` / `false`, `on` / `off`
`image_urls`	First-frame / last-frame URLs, maximum 2 URLs
`image_tail` / `last_frame_url`	Last-frame URL; takes precedence over `image_urls[1]`
`image_list`	Multi-reference images, maximum 9 images
`multi_shots` / `multi_shot`	Multi-shot toggle
`shot_type`	Shot type
`multi_prompt`	Array of multi-shot prompts
`kling_elements` / `element_list`	Kling element controls
`camera_control`	Camera movement control
`motion_brush` / `dynamic_masks`	Motion-control inputs
`video_list`	Video references, often used by O1 or video-input workflows
`voice_list`	Voice references
`session_id` / `face_choose`	Lip-sync face selection fields

Text-to-Video

cURL

curl -X POST https://api.crazyrouter.com/v1/video/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "aigc-video-kling-2.6",
    "prompt": "a red square logo slowly rotating on a clean white background",
    "size": "1280x720",
    "seconds": "5",
    "metadata": {
      "resolution": "720P",
      "sound": false
    }
  }'

Image-to-Video

cURL

curl -X POST https://api.crazyrouter.com/v1/video/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "aigc-video-kling-2.5-turbo",
    "prompt": "make the picture gently move with a slow camera push in",
    "image": "https://example.com/input.png",
    "size": "1280x720",
    "seconds": "5",
    "metadata": {
      "resolution": "720P"
    }
  }'

First-Last Frame Video

cURL

curl -X POST https://api.crazyrouter.com/v1/video/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "aigc-video-kling-2.6",
    "prompt": "transition naturally from the first frame to the last frame",
    "images": [
      "https://example.com/first.png",
      "https://example.com/last.png"
    ],
    "seconds": "5",
    "metadata": {
      "resolution": "720P",
      "sound": false
    }
  }'

For Kling 2.6 first-last frame requests, explicitly setting metadata.sound=false is recommended.

Multi-Reference Images

cURL

curl -X POST https://api.crazyrouter.com/v1/video/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "aigc-video-kling-3.0",
    "prompt": "keep the same character identity and create a cinematic walking shot",
    "seconds": "5",
    "metadata": {
      "resolution": "720P",
      "image_list": [
        "https://example.com/ref-1.png",
        "https://example.com/ref-2.png"
      ]
    }
  }'

Motion Control

Use a dedicated motion-control model such as aigc-video-kling-2.6-motion-control or aigc-video-kling-3.0-motion-control.

cURL

curl -X POST https://api.crazyrouter.com/v1/video/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "aigc-video-kling-3.0-motion-control",
    "prompt": "animate the selected subject moving from left to right",
    "image": "https://example.com/input.png",
    "seconds": "5",
    "metadata": {
      "resolution": "720P",
      "dynamic_masks": [
        {
          "mask_url": "https://example.com/mask.png",
          "trajectories": [
            { "x": 220, "y": 380 },
            { "x": 620, "y": 380 }
          ]
        }
      ]
    }
  }'

O1 Video Input

aigc-video-kling-o1 accepts video references through metadata.video_list. Pricing can differ depending on whether video input is present.

cURL

curl -X POST https://api.crazyrouter.com/v1/video/generations \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "aigc-video-kling-o1",
    "prompt": "use the reference video motion style and generate a new cinematic shot",
    "seconds": "5",
    "metadata": {
      "resolution": "720P",
      "video_list": [
        { "url": "https://example.com/reference.mp4" }
      ]
    }
  }'

Avatar and Lip Sync

Avatar requests use aigc-video-kling-avatar:

{
  "model": "aigc-video-kling-avatar",
  "prompt": "a presenter speaks naturally to the camera",
  "image": "https://example.com/avatar.png",
  "seconds": "5",
  "metadata": {
    "resolution": "720P",
    "voice_list": [
      { "url": "https://example.com/voice.wav" }
    ]
  }
}

Lip sync uses aigc-video-kling-identifyface. If your workflow already has upstream session_id and face_choose values, pass them in metadata:

{
  "model": "aigc-video-kling-identifyface",
  "prompt": "lip sync the face with the provided voice",
  "image": "https://example.com/face.png",
  "seconds": "5",
  "metadata": {
    "resolution": "720P",
    "session_id": "SESSION_ID",
    "face_choose": "FACE_ID",
    "voice_list": [
      { "url": "https://example.com/voice.wav" }
    ]
  }
}

A full lip-sync workflow usually requires face identification before video generation. Without session_id / face_choose, the upstream may return a parameter error.

Query a Task

After creation returns id or task_id, query the same compatibility route:

cURL

curl https://api.crazyrouter.com/v1/video/generations/TASK_ID \
  -H "Authorization: Bearer YOUR_API_KEY"

Completed tasks usually return an archived video URL:

{
  "code": "success",
  "data": {
    "status": "SUCCESS",
    "task_id": "vod_task_abc123",
    "result_url": "https://media.crazyrouter.com/task-artifacts/example.mp4",
    "artifact_url": "https://media.crazyrouter.com/task-artifacts/example.mp4"
  }
}

Differences from Native Kling

Item	AIGC Kling VOD	Native Kling
Model names	`aigc-video-kling-*`	`kling-v2-5-turbo`, `kling-v2-6`, `kling-v3`
Create route	`POST /v1/video/generations`	`POST /kling/v1/videos/text2video`, etc.
Query route	`GET /v1/video/generations/{task_id}`	`GET /kling/v1/videos/{type}/{task_id}`
Main fields	`model`, `prompt`, `seconds`, `image`, `images`, `metadata`	`model_name`, `prompt`, `image_urls`, `duration`, etc.
Best for	Tencent VOD route, VOD pricing and capabilities	Native Kling protocol

Use the live Pricing page as the source of truth for prices, discounts, and enabled specs.

Chat - OpenAI

Chat - Claude

Chat - Gemini

Chat - Responses API

Image Generation

Video Generation

Audio

Embeddings & Rerank

Other APIs

Reference

AIGC Kling VOD Guide

AIGC Kling VOD Guide

Available Models

Create a Task

Common Parameters

Common `metadata` Fields

Text-to-Video

Image-to-Video

First-Last Frame Video

Multi-Reference Images

Motion Control

O1 Video Input

Avatar and Lip Sync

Query a Task

Differences from Native Kling

​AIGC Kling VOD Guide

​Available Models

​Create a Task

​Common Parameters

​Common metadata Fields

​Text-to-Video

​Image-to-Video

​First-Last Frame Video

​Multi-Reference Images

​Motion Control

​O1 Video Input

​Avatar and Lip Sync

​Query a Task

​Differences from Native Kling

AIGC Kling VOD Guide

Available Models

Create a Task

Common Parameters

Common `metadata` Fields

Text-to-Video

Image-to-Video

First-Last Frame Video

Multi-Reference Images

Motion Control

O1 Video Input

Avatar and Lip Sync

Query a Task

Differences from Native Kling