Kuaishou’s video model. Your frames, your characters, your voices.
Kling 4.0 AI video generator
Up to 30 seconds a shot. Native sound. Up to 1080p.
Real takes. Real prompts.
What Kling 4.0 does best
Start with your own frame
Bring a still image and describe the motion, action and camera move you want in the next take.
More time for the camera
Build a shot up to 30 seconds long, or set start and end frames to guide where it begins and finishes.
Sound in the same take
Write spoken lines in the prompt. For recurring characters, attach up to 3 actor voices in supported workflows.
At a glance
Kling 4.0 by the numbers
30s
Longest clip
1080p
Top resolution
10 refs
Image references in one generation
3 voices
Actor voice references in supported workflows
Audio
Dialogue and sound with the picture
2 frames
Guide the start and the end
- Made by
- Kuaishou
- Clip length
- 3 to 30 seconds
- Resolution
- 720p and 1080p
- References
- Up to 10 images and 3 character voices
- Frames
- Start and end frames
- Sound
- Native dialogue and sound
- Flash
- Up to 20 seconds at 720p, without voice references
- In PPS
- Director Mode, Storyboard Mode, Shorts and Image Studio
Prompt showcase
The prompts behind these
Same frame. Every model.
One image. Five models.
A runner in the rain
Prompt: A woman sprints through heavy rain down a neon-lit city street at night, splashing through puddles. The camera tracks backward in front of her at running speed. Rain streaks through the light, her breath fogs, and a car's headlights sweep past behind her.
A car at night
Prompt: A vintage black car drives slowly along a wet city street at night. The camera glides low beside the front wheel, then rises to reveal the driver through the side window. Neon signs reflect and ripple across the wet paint and the road.
A close-up with a spoken line
Prompt: Close-up of a woman looking into the camera. She smiles slightly and says, "You only get one first impression, so make it count." Soft window light, shallow depth of field, a slow push-in.
How it works
From idea to finished film
- 1
Start with a script or an idea
Paste a script, or describe the video you want. The studio breaks it into scenes and shots.
- 2
Lock your characters once
Create an Actor from a photo or a description, and the same face is reused in every shot.
- 3
Pick a model for each shot
Use Seedance, Kling or Veo shot by shot, and see the credit cost before you generate.
- 4
Edit and export
Arrange your takes on the timeline, add voices and music, and export an MP4 or a Premiere Pro project.
Pricing
What Kling 4.0 costs
203
credits for a 10-second clip at 720p
305
credits for a 10-second clip at 1080p
In Pray Studio, a 10-second Kling 4.0 clip uses about 203 credits at 720p and 305 credits at 1080p. Plans start at $19 a month with 1,000 credits, and every Generate button shows the exact cost before you run it.
Questions, answered
What is Kling 4.0?
Kling 4.0 is Kuaishou’s video generation model. It turns a prompt and optional images into a 3 to 30 second clip with native sound, at 720p and 1080p.
How is Kling 4.0 different from Kling 3.0?
Kling 4.0 runs up to 30 seconds, compared with 15 seconds for Kling 3.0. It also takes up to 10 image references and 3 character voices. Kling 3.0 costs less per second.
Can Kling 4.0 use a start and end frame?
Yes. Set the first and last images to guide the shot between them. Reference images can also guide the characters and setting.
What is Kling 4.0 Flash?
Flash is the lower-cost option for clips up to 20 seconds at 720p. It accepts a start image and image references, but not end frames or character voice references.
Can I mix Kling 4.0 with other video models?
Yes. Choose Kling, Seedance or Veo shot by shot and arrange their clips on the same timeline in Pray Studio.
How much does Kling 4.0 cost in Pray Studio?
In Pray Studio, a 10-second Kling 4.0 clip uses about 203 credits at 720p and 305 credits at 1080p. Plans start at $19 a month with 1,000 credits, and every Generate button shows the exact cost before you run it.
Kling 4.0
Write a script or an idea, lock your characters once, and generate shot by shot with the model that fits each shot.