Skip to main content

kling/kling-v4-omni-reference-video-generation

Supports text-to-video, keyframe-to-video, reference-to-video, and video editing. Always generates video with sound.

Inference service provider

The inference service provider for the kling/kling-v4-omni-reference-video-generation model is Kling AI.

Model capabilities

Input modalitiesOutput modalitySupported capabilities
Image, Text, VideoVideoText-to-video, keyframe-to-video, reference-to-video, and video editing
The model always generates video with sound. It supports std (720P) and pro (1080P), generates 3–30 second videos, and outputs MP4 (H.264). For usage, see Kling - Video Generation API.

Pricing

This page shows the original model invocation prices only and does not include limited-time promotions. Go to the Model Studio console to view promotional offers.
  • China (Beijing)
Billing itemPrice (CNY)Unit
Video generation (720P, without reference video)1.0Per second
Video generation (1080P, without reference video)1.5Per second
Video generation (720P, with reference video)0.6Per second
Video generation (1080P, with reference video)0.9Per second
Without video input, the billable duration equals the generated video duration. With video input, it equals max(total input video duration, generated video duration) + generated video duration. Free quota: None. Failed requests are not billed.

Rate limits

  • China (Beijing)
ParameterValue
Requests per second (RPS)5
Concurrent processing tasks10 (shared)
Under the same Model Studio API key, seven Kling image and video models share a concurrency limit of 10. For details, see Model rate limits.