Skip to main content
POST
Kling v3.0 4K Image-to-Video
Kling v3.0 4K Image-to-Video generates native 4K ultra-high-definition videos from a first-frame image with cinematic quality, smooth motion, precise prompt adherence, and optional synchronized audio generation. Supports end-frame guidance, multi-shot composition, and flexible 3-15 second durations.
This is an asynchronous API; only the task_id will be returned. You should use the task_id to request the Task Result API to retrieve the video generation results.

Request Headers

string
required
Supports: application/json
string
required
Bearer authentication format, for example: Bearer {{API Key}}.

Request Body

string
required
First frame image for video; supports .jpg, .jpeg, .png. Image file size must not exceed 10MB; width and height must be >= 300px; aspect ratio must be between 1:2.5 and 2.5:1.
boolean
default:false
Whether to generate synchronized audio simultaneously with the video. Supports Chinese and English voice output.
string
required
Positive prompt text for video generation, describing scene motion, camera movement, actions, voice style, atmosphere, and sound effects; must not exceed 2500 characters.Length limit: 0 - 2500
integer
default:5
Duration of generated video in seconds. Supports flexible durations from 3 to 15 seconds.Value range: [3, 15]
number
Controls flexibility of video generation. Higher values result in content more closely following the prompt; lower values produce more natural motion.Value range: [0, 1]
string
Ending frame image URL for guided transitions between start and end frames. Same format constraints as image. Cannot be used together with multi_prompt.
object[]
Array of prompts for multi-shot video composition. Each item contains a prompt and duration for one segment. Cannot be used together with end_image.
string
Negative prompt describing elements to avoid in video and audio; must not exceed 2500 characters.Length limit: 0 - 2500

Response

string
required
Use the task_id to request the Task Result API to retrieve the generated outputs.
Last modified on July 3, 2026