Kling Omni Video

Kling Omni Video is Kling's unified multimodal video generation endpoint, supporting rich multimodal inputs including multiple images, video references, and subject control, enabling precise motion control and rich visual expression.

V3 New Features

The kling-v3-omni model introduces the following new capabilities on the Omni endpoint:

• Extended Duration: Supports 3-15 seconds (legacy O1 supports 3-10 seconds)
• Multi-shot Narrative: Generate multiple consecutive shots in a single request, up to 6 shots
• Video Editing Mode: Edit existing videos with text instructions via refer_type: "base"
• Native Audio: Supports generating synchronized audio (with multilingual lip sync)
• VIDEO Element Reference: element_list supports character elements extracted from video clips

API Details

Endpoint: POST /kling/v1/videos/omni-video

Description: Submit an Omni video generation task. Supports mixed inputs (text, images, videos, subject elements).

Authentication: Bearer Token

Authorization: Bearer YOUR_API_TOKEN

---

Supported Models

Model	Description	Duration Range	Multi-shot	Video Editing	Native Audio
kling-v3-omni	Video 3.0, latest version	3-15s	✅	✅	✅
kling-video-o1	Omni V1	3-10s	❌	❌	❌

---

Request Parameters

Body Parameters

Name	Type	Required	Default	Description	Example
model	string	Yes	-	Model ID	kling-v3-omni
prompt	string	Conditional	-	Video description text. Cannot be used together with multi_prompt	A small fox running in the forest
negative_prompt	string	No	-	Negative prompts	blur, watermark
mode	string	No	std	Generation mode: std (Standard 720p), pro (Professional 1080p)	std, pro
duration	string	No	5	Video duration (seconds). Cannot be used together with multi_prompt. See Duration Rules below	5, 10, 15
aspect_ratio	string	No	16:9	Video aspect ratio	16:9, 9:16, 1:1
image_list	array	No	-	Reference image list	See OmniImageItem below
video_list	array	No	-	Reference video list (using video reference increases cost multiplier)	See OmniVideoItem below
element_list	array	No	-	Subject element list	[{"element_id": 123456}]
multi_prompt	array	No	-	V3 Only Multi-shot scene list, up to 6 shots	See MultiShotItem below
sound	string	No	off	Whether to generate native audio. V3 supports on/off. Note: Cannot be on when using video_list	on, off
cfg_scale	float	No	0.5	Prompt adherence scale	0.0 - 1.0
external_task_id	string	No	-	Custom task ID	my_task_001
callback_url	string	No	-	Callback URL after task completion	https://your-api.com/callback

Duration Rules

Model	Workflow	Supported Duration
kling-v3-omni	T2V / I2V	3 - 15
kling-v3-omni	Video reference (refer_type=feature)	3 - 10
kling-v3-omni	Multi-shot	Sum of all shot durations must be within 3-15 range
kling-v3-omni	Video editing (refer_type=base)	Automatically follows original video duration
kling-video-o1	T2V / basic I2V	5, 10
kling-video-o1	First/end frame / video reference	3 - 10

OmniImageItem

Name	Type	Description
image_url	string	Image URL
type	string	Image role: first_frame, end_frame. Omit for general reference image

OmniVideoItem

Name	Type	Description
video_url	string	Video URL
refer_type	string	Reference type: feature (feature reference) or base (video editing, kling-v3-omni only)
keep_original_sound	string	Keep original sound: yes, no

MultiShotItem (V3 Only)

Defines a single shot in multi-shot mode. When using multi_prompt, the top-level prompt and duration parameters cannot be used.

Name	Type	Required	Description
prompt	string	Yes	Description text for this shot, max 2500 characters. Supports @image_1, @element_1 reference syntax
duration	string	Yes	Duration of this shot (seconds), minimum 3 seconds per shot

Multi-shot Constraints

• Minimum 2 shots, maximum 6 shots
• Total duration of all shots must be within 3-15 seconds
• Each shot must be at least 3 seconds

Audio and Video Input are Mutually Exclusive

When using video_list (video reference or video editing), sound: "on" is not supported. The upstream API will return: sound on is not supported with video input.

---

Pricing

Model	Mode	Base Price	With Audio	With Video Input
kling-v3-omni	Std	Per-second billing	×1.5	×1.5
kling-v3-omni	Pro	Per-second billing ×1.333	×1.5	×1.5
kling-video-o1	Std	Per-second billing	-	×1.5
kling-video-o1	Pro	Per-second billing ×1.333	-	×1.5

---

Response Parameters

Returns an OpenAI-compatible video object upon successful submission.

Name	Type	Description
id	string	Task ID
task_id	string	Task ID (Compatibility field)
object	string	Object type, fixed as video
model	string	Model ID used
created_at	integer	Creation timestamp

---

Code Examples

Basic Usage - Text to Video

curl https://api.ezmodel.cloud/kling/v1/videos/omni-video \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $YOUR_API_KEY" \
  -d '{
    "model": "kling-v3-omni",
    "prompt": "A fox running through a sunlit forest with dappled light filtering through the leaves",
    "duration": "10",
    "aspect_ratio": "16:9",
    "sound": "on"
  }'

Image to Video - First/End Frame Mode

curl https://api.ezmodel.cloud/kling/v1/videos/omni-video \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $YOUR_API_KEY" \
  -d '{
    "model": "kling-v3-omni",
    "prompt": "The character slowly sits down from a standing position",
    "image_list": [
      {"image_url": "https://example.com/start.jpg", "type": "first_frame"},
      {"image_url": "https://example.com/end.jpg", "type": "end_frame"}
    ],
    "duration": "5"
  }'

V3 Multi-shot Narrative

curl https://api.ezmodel.cloud/kling/v1/videos/omni-video \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $YOUR_API_KEY" \
  -d '{
    "model": "kling-v3-omni",
    "multi_prompt": [
      {"prompt": "A girl pushes open the door of a coffee shop and walks in, camera follows", "duration": "4"},
      {"prompt": "The girl sits by the window, opens her laptop, close-up shot", "duration": "4"},
      {"prompt": "Rain starts falling outside, the girl smiles and looks out the window, medium shot", "duration": "5"}
    ],
    "aspect_ratio": "16:9",
    "mode": "pro",
    "sound": "on"
  }'

V3 Video Editing (refer_type=base)

curl https://api.ezmodel.cloud/kling/v1/videos/omni-video \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $YOUR_API_KEY" \
  -d '{
    "model": "kling-v3-omni",
    "prompt": "Replace the background with a snowy mountain landscape",
    "video_list": [
      {
        "video_url": "https://example.com/original.mp4",
        "refer_type": "base",
        "keep_original_sound": "yes"
      }
    ]
  }'

Response Example

{
  "id": "842250903629086785",
  "task_id": "842250903629086785",
  "object": "video",
  "model": "kling-v3-omni",
  "created_at": 1737367800
}