kling-v3 & kling-v3-omni

Kling Legacy APIs for Kling 3.0 and Kling 3.0 Omni.

Text to Video

Create Task

POST /kling/v1/videos/text2video

Generate video from text with model_name: "kling-3.0". Multi-shot storyboards, camera movement, sound, and voice references are supported.

cURL

curl --request POST \
  --url https://www.anyfast.ai/kling/v1/videos/text2video \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
    "model_name": "kling-3.0",
    "prompt": "A rabbit wearing glasses reads a newspaper at a table",
    "duration": "5",
    "mode": "pro",
    "sound": "on",
    "aspect_ratio": "1:1"
  }'

Create response

200

{
  "id": "860260753860210752",
  "task_id": "860260753860210752",
  "object": "video",
  "model": "kling-v3",
  "status": "queued",
  "progress": 0,
  "created_at": 1773130665
}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Data exchange format. Use application/json.

Request parameters

model_name · string · required

Use kling-3.0.

prompt · string

Positive prompt, up to 2,500 characters; required for a single shot or intelligent storyboard.

negative_prompt · string

Negative prompt, up to 2,500 characters.

multi_shot · boolean · default: false

Enable multi-shot generation.

shot_type · string

With multi-shot, use customize or intelligence.

multi_prompt · object[]

One to six custom shots whose durations add up to duration.

voice_list · object[]

Up to two voice references.

sound · string · default: "off"

Use on or off; voice references require on.

mode · string · default: "std"

std for 720p, pro for 1080p, or 4k.

duration · string · default: "5"

Video duration from 3 to 15 seconds.

aspect_ratio · string · default: "16:9"

Use 16:9, 9:16, or 1:1.

cfg_scale · number

Prompt adherence from 0 to 1.

camera_control · object

Preset or custom camera movement.

watermark_info · object

Platform watermark configuration through enabled.

Storyboards, voices, and camera movement

When multi_shot is true, set shot_type to customize and supply multi_prompt, or use intelligence with prompt. Each custom shot contains index, prompt, and duration; each shot lasts at least 1 second and its prompt is at most 512 characters.

Voice references use <<<voice_1>>> and <<<voice_2>>> in list order. Supply at most two voice_id values from voice management and set sound to on.

camera_control.type supports simple, down_back, forward_up, right_turn_forward, and left_turn_forward. For simple, only one of horizontal, vertical, pan, tilt, roll, and zoom may be nonzero; every value ranges from -10 to 10.

Interactive request parameters

Fill in and test the full Legacy request body.

Query task by ID

GET /kling/v1/videos/text2video/{task_id}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Data exchange format. Use application/json.

Path parameters

task_id · string · required

Text-to-video task ID returned by the create endpoint.

cURL

curl --request GET \
  --url https://www.anyfast.ai/kling/v1/videos/text2video/TASK_ID \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json'

200

{
  "code": 0,
  "data": {
    "created_at": 1784720894000,
    "task_id": "asyntask_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
    "task_result": {
      "videos": [{
        "duration": "5",
        "url": "https://example.com/generated-video.mp4"
      }]
    },
    "task_status": "succeed",
    "updated_at": 1784721103000
  },
  "message": ""
}

Response fields

code · integer

0 indicates that the query was accepted.

message · string

Error information; empty for a successful query.

data.task_id · string

Queried task ID.

data.task_status · string

submitted, processing, succeed, or failed.

data.task_status_msg · string

Failure details when the task fails.

data.task_result.videos · object[]

Generated videos, returned after success.

data.task_result.videos[].url · string

Temporary generated-video URL.

data.task_result.videos[].duration · string

Actual duration in seconds.

data.created_at · integer

Creation Unix timestamp in milliseconds.

data.updated_at · integer

Last update Unix timestamp in milliseconds.

Image to Video

Create Task

POST /kling/v1/videos/image2video

Generate video from a first frame, an end frame, or both with model_name: "kling-3.0".

cURL

curl --request POST \
  --url https://www.anyfast.ai/kling/v1/videos/image2video \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
    "model_name": "kling-3.0",
    "image": "https://example.com/first-frame.png",
    "prompt": "The character turns and walks toward the window",
    "duration": "5",
    "mode": "pro",
    "sound": "off"
  }'

200

{"id":"860260753860210752","task_id":"860260753860210752","object":"video","model":"kling-v3","status":"queued","progress":0,"created_at":1773130665}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Data exchange format. Use application/json.

Request parameters

model_name · string · required

Use kling-3.0.

image · string

First frame as URL or raw Base64; at least one frame is required.

image_tail · string

End frame; mutually exclusive with motion brush and camera control.

prompt · string

Positive prompt, up to 2,500 characters.

negative_prompt · string

Negative prompt, up to 2,500 characters.

multi_shot · boolean · default: false

Enable multi-shot generation.

shot_type · string

Use customize or intelligence.

multi_prompt · object[]

One to six custom shots.

element_list · object[]

Up to three elements; mutually exclusive with voice_list.

voice_list · object[]

Up to two voices; requires sound: "on".

static_mask · string

Static motion-brush mask.

dynamic_masks · object[]

Up to six dynamic masks.

camera_control · object

Preset or simple camera movement.

mode · string · default: "std"

Use std, pro, or 4k.

duration · string · default: "5"

Video duration from 3 to 15 seconds.

cfg_scale · number

Prompt adherence from 0 to 1.

watermark_info · object

Platform watermark setting.

multi_prompt[].index · integer · required

Shot index, starting from 1.

multi_prompt[].prompt · string · required

Shot prompt, up to 512 characters.

multi_prompt[].duration · string · required

Shot duration, at least 1 second; the sum must equal top-level duration.

element_list[].element_id · integer · required

Element ID returned by element management.

voice_list[].voice_id · string · required

Voice ID returned by voice management.

dynamic_masks[].mask · string · required

Dynamic mask URL or raw Base64.

dynamic_masks[].trajectories · object[] · required

Motion trajectory coordinates.

dynamic_masks[].trajectories[].x · integer · required

Horizontal coordinate from the bottom-left origin.

dynamic_masks[].trajectories[].y · integer · required

Vertical coordinate from the bottom-left origin.

camera_control.type · string · required

simple or one of four preset compound movements.

camera_control.config · object

Simple movement configuration; only one of six values may be nonzero.

watermark_info.enabled · boolean · default: false

Add the platform watermark.

Create response

id · string

AnyFast task ID.

task_id · string

Task ID, equal to id.

object · string

Always video.

model · string

Model used by the task.

status · string

Initial task status.

progress · integer

Initial progress.

created_at · integer

Creation Unix timestamp.

Image, storyboard, and motion-brush rules

image and image_tail accept a public URL or raw Base64 without a data URI prefix. Use JPG, JPEG, or PNG up to 10 MB, at least 300 px per side, with an aspect ratio from 1:2.5 to 2.5:1. At least one frame is required. An end frame, motion brush, and camera control are mutually exclusive.

Dynamic motion brush supports up to 6 masks. Mask dimensions must match the input image and each other. For a 5-second video, each trajectory contains 2–77 coordinates measured from the bottom-left corner.

Multi-shot, voice, and camera-control rules are the same as text-to-video. element_list and voice_list cannot be used together.

Interactive request parameters

Fill in and test every first/end-frame, storyboard, brush, element, voice, and camera field.

Query task by ID

GET /kling/v1/videos/image2video/{task_id}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Data exchange format. Use application/json.

Path parameters

task_id · string · required

Image-to-video task ID returned by the create endpoint.

cURL

curl --request GET \
  --url https://www.anyfast.ai/kling/v1/videos/image2video/TASK_ID \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json'

200

{
  "code": 0,
  "data": {
    "created_at": 1784720894000,
    "task_id": "asyntask_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
    "task_result": {
      "videos": [{
        "duration": "5",
        "url": "https://example.com/generated-video.mp4"
      }]
    },
    "task_status": "succeed",
    "updated_at": 1784721103000
  },
  "message": ""
}

Response fields

code · integer

0 indicates that the query was accepted.

message · string

Error information; empty for a successful query.

data.task_id · string

Queried task ID.

data.task_status · string

submitted, processing, succeed, or failed.

data.task_status_msg · string

Failure details when the task fails.

data.task_result.videos · object[]

Generated videos, returned after success.

data.task_result.videos[].url · string

Temporary generated-video URL.

data.task_result.videos[].duration · string

Actual duration in seconds.

data.created_at · integer

Creation Unix timestamp in milliseconds.

data.updated_at · integer

Last update Unix timestamp in milliseconds.

Omni Video Generation

Create Task

POST /kling/v1/videos/omni-video

Use model_name: "kling-3.0-omni" with text, images, video, and elements for generation or editing.

cURL

curl --request POST \
  --url https://www.anyfast.ai/kling/v1/videos/omni-video \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
    "model_name": "kling-3.0-omni",
    "prompt": "The character in <<<image_1>>> walks into a rainy neon street",
    "image_list": [{"image_url": "https://example.com/character.png"}],
    "duration": "5",
    "mode": "pro",
    "aspect_ratio": "16:9"
  }'

200

{"id":"860260753860210752","task_id":"860260753860210752","object":"video","model":"kling-v3-omni","status":"queued","progress":0,"created_at":1773130665}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Data exchange format. Use application/json.

Request parameters

model_name · string · required

Use kling-3.0-omni.

prompt · string

Up to 2,500 characters; reference media with the official placeholders.

image_list · object[]

Reference images and optional first/end frames.

video_list · object[]

At most one reference video, including refer_type and keep_original_sound.

element_list · object[]

Element count depends on whether a video is supplied.

multi_shot · boolean · default: false

Enable multi-shot generation.

shot_type · string

Use customize or intelligence.

multi_prompt · object[]

One to six custom shots.

mode · string · default: "pro"

Use std, pro, or 4k.

aspect_ratio · string

Use 16:9, 9:16, or 1:1 when applicable.

duration · string · default: "5"

3–15 seconds; ignored for refer_type: "base".

sound · string · default: "off"

Must be off when video_list is present.

watermark_info · object

Platform watermark setting.

image_list[].image_url · string · required

Reference image URL or raw Base64.

image_list[].type · string

first_frame or end_frame; an end frame requires a first frame.

video_list[].video_url · string · required

Reference video URL.

video_list[].refer_type · string · default: "base"

base edits the source video; feature references its characteristics.

video_list[].keep_original_sound · string

Use yes or no.

element_list[].element_id · integer · required

Element ID from element management.

multi_prompt[].index · integer · required

Shot index.

multi_prompt[].prompt · string · required

Shot prompt, up to 512 characters.

multi_prompt[].duration · string · required

Shot duration, at least 1 second.

watermark_info.enabled · boolean · default: false

Add the platform watermark.

Create response

id · string

AnyFast task ID.

task_id · string

Task ID, equal to id.

object · string

Always video.

model · string

kling-v3-omni.

status · string

Initial task status.

progress · integer

Initial progress.

created_at · integer

Creation Unix timestamp.

Media constraints

Images accept URL or raw Base64, JPG/JPEG/PNG, up to 10 MB, at least 300 px per side, and aspect ratio 1:2.5 to 2.5:1. An end frame requires a first frame. Without a reference video, images plus multi-image elements total at most 7 and video-character elements at most 3. With a reference video, those limits become 4 and 1.

Reference video accepts MP4/MOV, 3–15.5 seconds, up to 200 MB, 24–60 fps, dimensions 700–4553 px, no more than 8,294,400 pixels, and aspect ratio 1:2.5 to 2.5:1.

refer_type: "base" edits the original video, preserves its duration, and disables multi-shot and first/end-frame controls. refer_type: "feature" references content, motion, style, or camera movement and can use intelligent multi-shot. keep_original_sound controls source audio; new sound generation is unavailable whenever a reference video is present.

Interactive request parameters

Fill in and test the complete Omni Legacy request body.

Query task by ID

GET /kling/v1/videos/omni-video/{task_id}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Data exchange format. Use application/json.

Path parameters

task_id · string · required

Omni-video task ID returned by the create endpoint.

cURL

curl --request GET \
  --url https://www.anyfast.ai/kling/v1/videos/omni-video/TASK_ID \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json'

200

{
  "code": 0,
  "data": {
    "created_at": 1784516360000,
    "task_id": "asyntask_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
    "task_result": {
      "videos": [{
        "duration": "5",
        "url": "https://example.com/generated-video.mp4"
      }]
    },
    "task_status": "succeed",
    "updated_at": 1784516573000
  },
  "message": ""
}

Response fields

code · integer

0 indicates that the query was accepted.

message · string

Error information; empty for a successful query.

data.task_id · string

Queried task ID.

data.task_status · string

submitted, processing, succeed, or failed.

data.task_status_msg · string

Failure details when the task fails.

data.task_result.videos · object[]

Generated videos, returned after success.

data.task_result.videos[].url · string

Temporary generated-video URL.

data.task_result.videos[].duration · string

Actual duration in seconds.

data.created_at · integer

Creation Unix timestamp in milliseconds.

data.updated_at · integer

Last update Unix timestamp in milliseconds.

Motion Control

Create Task

POST /kling/v1/videos/motion-control

Transfer movement from a reference video to a character image with model_name: "kling-3.0".

cURL

curl --request POST \
  --url https://www.anyfast.ai/kling/v1/videos/motion-control \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
    "model_name": "kling-3.0",
    "image_url": "https://example.com/character.png",
    "video_url": "https://example.com/motion.mp4",
    "prompt": "The character performs the reference action in a studio",
    "keep_original_sound": "yes",
    "character_orientation": "image",
    "mode": "pro"
  }'

200

{"id":"860260753860210752","task_id":"860260753860210752","object":"video","model":"kling-v3","status":"queued","progress":0,"created_at":1773130665}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Data exchange format. Use application/json.

Request parameters

model_name · string · required

Use kling-3.0.

image_url · string · required

Character image as URL or raw Base64.

video_url · string · required

Motion reference video URL.

prompt · string

Scene, motion, or camera guidance, up to 2,500 characters.

element_list · object[]

At most one element.

keep_original_sound · string · default: "yes"

Use yes or no.

character_orientation · string · required

Use image or video.

mode · string · required

Use std or pro.

watermark_info · object

Platform watermark setting.

element_list[].element_id · integer · required

Element ID from element management; at most one.

watermark_info.enabled · boolean · default: false

Add the platform watermark.

Create response

id · string

AnyFast task ID.

task_id · string

Task ID, equal to id.

object · string

Always video.

model · string

kling-v3.

status · string

Initial task status.

progress · integer

Initial progress.

created_at · integer

Creation Unix timestamp.

The character image must be JPG/JPEG/PNG up to 10 MB, dimensions 300–65,536 px, and aspect ratio 1:2.5 to 2.5:1. The MP4/MOV motion video can be up to 100 MB with dimensions 340–3850 px. With character_orientation: "image", the video may be 3–10 seconds; with video, it may be 3–30 seconds. Element references follow the video’s orientation.

Complex or fast motion can produce a result shorter than the uploaded video because the model only extracts valid continuous action segments. Generation can proceed once at least 3 seconds of usable continuous motion is found. Credits consumed in this case are non-refundable, so reduce the motion complexity or speed when necessary.

Interactive request parameters

Fill in and test the complete motion-control request body.

Query task by ID

GET /kling/v1/videos/motion-control/{task_id}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Data exchange format. Use application/json.

Path parameters

task_id · string · required

Motion-control task ID returned by the create endpoint.

cURL

curl --request GET \
  --url https://www.anyfast.ai/kling/v1/videos/motion-control/TASK_ID \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json'

200

{
  "code": 0,
  "data": {
    "created_at": 1784720894000,
    "task_id": "asyntask_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx",
    "task_result": {
      "videos": [{
        "duration": "5",
        "url": "https://example.com/generated-video.mp4"
      }]
    },
    "task_status": "succeed",
    "updated_at": 1784721103000
  },
  "message": ""
}

Response fields

code · integer

0 indicates that the query was accepted.

message · string

Error information; empty for a successful query.

data.task_id · string

Queried task ID.

data.task_status · string

submitted, processing, succeed, or failed.

data.task_status_msg · string

Failure details when the task fails.

data.task_result.videos · object[]

Generated videos, returned after success.

data.task_result.videos[].url · string

Temporary generated-video URL.

data.task_result.videos[].duration · string

Actual duration in seconds.

data.created_at · integer

Creation Unix timestamp in milliseconds.

data.updated_at · integer

Last update Unix timestamp in milliseconds.

Element Management

Element APIs

Operation	Method and path
Create custom element	POST /kling/v1/general/advanced-custom-elements
Query custom element	GET /kling/v1/general/advanced-custom-elements/{task_id}
List preset elements	GET /kling/v1/general/advanced-presets-elements

Create a custom element

cURL

curl --request POST \
  --url https://www.anyfast.ai/kling/v1/general/advanced-custom-elements \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
    "element_name": "my-character",
    "element_description": "The lead character",
    "reference_type": "image_refer",
    "element_image_list": {
      "frontal_image": "https://example.com/front.png",
      "refer_images": [
        {"image_url": "https://example.com/side.png"}
      ]
    },
    "tag_list": [{"tag_id": "o_102"}]
  }'

200

{
  "id": "863121016086724692",
  "task_id": "863121016086724692",
  "object": "video",
  "model": "kling-element",
  "status": "queued",
  "progress": 0,
  "created_at": 1773812605
}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Use application/json.

Request body

element_name · string · required

Element name, up to 20 characters.

element_description · string · required

Element description, up to 100 characters.

reference_type · string · required

image_refer for image references or video_refer for one reference video.

element_image_list · object

Required when reference_type is image_refer.

element_image_list.frontal_image · string

Front-view image URL or raw Base64 without a data URI prefix.

element_image_list.refer_images · object[]

One to three alternate-angle or close-up images.

element_image_list.refer_images[].image_url · string

Reference image URL or raw Base64.

element_video_list · object

Required when reference_type is video_refer.

element_video_list.refer_videos · object[]

Exactly one reference video.

element_video_list.refer_videos[].video_url · string

Public MP4 or MOV URL. The video must be 3–8 seconds, 1080p, 16:9 or 9:16, and no larger than 200 MB.

element_voice_id · string

Optional voice ID to bind to the element.

tag_list · object[]

Element tags.

tag_list[].tag_id · string · required

o_101 trend, o_102 character, o_103 animal, o_104 item, o_105 costume, o_106 scene, o_107 effect, or o_108 other.

Images support JPG, JPEG, and PNG, up to 10 MB each. Both dimensions must be at least 300 px and the aspect ratio must be between 1:2.5 and 2.5:1.

Create response

id · string

AnyFast task ID.

task_id · string

Task ID, identical to id.

object · string

Object type; currently video.

model · string

Fixed value kling-element.

status · string

Initial task status, normally queued.

progress · integer

Task progress from 0 to 100.

created_at · integer

Creation Unix timestamp.

Query a custom element

GET /kling/v1/general/advanced-custom-elements/{task_id}

cURL

curl --request GET \
  --url https://www.anyfast.ai/kling/v1/general/advanced-custom-elements/TASK_ID \
  --header 'Authorization: Bearer YOUR_API_KEY'

task_id · string · required

Task ID returned by the create endpoint.

The result uses data.task_status (submitted, processing, succeed, or failed). After success, read data.task_result.elements[]; each item includes element_id, name, description, reference media, optional voice information, tags, ownership, and status.

List preset elements

GET /kling/v1/general/advanced-presets-elements?pageNum=1&pageSize=30

cURL

curl --request GET \
  --url 'https://www.anyfast.ai/kling/v1/general/advanced-presets-elements?pageNum=1&pageSize=30' \
  --header 'Authorization: Bearer YOUR_API_KEY'

pageNum · integer · default: 1

Page number from 1 to 1,000.

pageSize · integer · default: 30

Items per page from 1 to 500.

The response returns task records in data[]. Read each record’s preset elements from task_result.elements[]. Current preset elements return owned_by: "vod".

This page strictly documents the Kling Legacy element API.

Voice Management

Voice APIs

Operation	Method and path
Create custom voice	POST /kling/v1/general/custom-voices
Query custom voice	GET /kling/v1/general/custom-voices/{id}
List preset voices	GET /kling/v1/general/presets-voices

Create and query

voice_name is at most 20 characters. Supply at least one of voice_url or video_id. A voice URL may point to MP3, WAV, MP4, or MOV containing one clean speaker for 5–30 seconds. A qualifying historical generated video can be supplied through video_id under the same audio constraints.

cURL

curl --request POST \
  --url https://www.anyfast.ai/kling/v1/general/custom-voices \
  --header 'Authorization: Bearer YOUR_API_KEY' \
  --header 'Content-Type: application/json' \
  --data '{
    "voice_name": "narrator",
    "voice_url": "https://example.com/voice.mp3"
  }'

200

{"id":"850087542757535834","task_id":"850087542757535834","object":"video","model":"kling-voices","status":"queued","progress":0,"created_at":1773812605}

Request headers

Authorization · string · required

Bearer authentication in the form Bearer YOUR_API_KEY.

Content-Type · string · default: "application/json" · required

Data exchange format. Use application/json.

Request parameters

voice_name · string · required

Custom voice name, up to 20 characters.

voice_url · string

Public MP3, WAV, MP4, or MOV URL; at least one source is required.

video_id · string

Eligible historical generated-video ID; at least one source is required.

Create response

id · string

AnyFast task ID.

task_id · string

Task ID, identical to id.

object · string

Object type; currently video.

model · string

Fixed value kling-voices.

status · string

Initial task status, normally queued.

progress · integer

Task progress from 0 to 100.

created_at · integer

Creation Unix timestamp.

Query custom voice

GET /kling/v1/general/custom-voices/{id}

cURL

curl --request GET \
  --url https://www.anyfast.ai/kling/v1/general/custom-voices/TASK_ID \
  --header 'Authorization: Bearer YOUR_API_KEY'

List preset voices

GET /kling/v1/general/presets-voices?pageNum=1&pageSize=30

cURL

curl --request GET \
  --url 'https://www.anyfast.ai/kling/v1/general/presets-voices?pageNum=1&pageSize=30' \
  --header 'Authorization: Bearer YOUR_API_KEY'

Voice Management APIs

Open the interactive reference for each voice operation.

These references use Kling’s Legacy request structures. AnyFast adds the /kling route prefix while retaining the Legacy endpoint names and request fields.