Create a stylized 3D BMX sequence beneath a graffiti-covered city overpass. A pink-haired rider in a green sweatshirt and loose cream trousers races up a ramp on a teal bike, launches into an aerial trick, and lands smoothly. Follow the action with overhead views, close tracking shots, and dramatic low angles. Keep the rider and bike consistent, with believable momentum, warm afternoon light, and crisp animated textures.
Made with Kling 4.0
One idea, three speeds
How the new
controls work
One take,
thirty seconds
Ten keyframes,
not two
Edit the shot,
not the prompt
Everything 4.0 adds
Shots that
used to break
How to generate
From references to final cut
Input Image Reference
Upload any reference image or video to guide your vision
Write the prompt
Use natural language to describe the desired scenario and sounds
Generate and export
Click the “Generate” button and receive a high-fidelity video in seconds
Community over 40 MILLION USERS
Join a global creative network where people generate AI images, share ideas, and inspire each other every day.




Trusted by 40M+ users worldwide
Still have questions?
We’ve answered the most frequently asked questions
What is Kling 4.0?
Kling AI’s latest video model. It generates up to 30 seconds in one pass, follows up to 10 keyframes and takes up to 15 references in a single task: up to 10 images and 5 videos, with saved subjects (up to 7) counting toward the 15.
How is it different from Kling 3.0?
3.0 stops at 15 seconds and only lets you set a start and an end frame. 4.0 doubles the length, adds multi-keyframe control, a 15-reference system, stereo audio, 21:9 output and in-place video editing.
What is Kling 4.0 Flash?
A faster, lighter version of the same model, up to 720p and 20 seconds. Use it to test a prompt or a camera move before rendering with the full model.
How long can a video be?
3 to 30 seconds in one generation. Upload an existing clip and you can extend it up to 2 minutes.
Which languages does it speak?
English, Spanish, Portuguese, German, French, Hindi, Japanese, Korean and Chinese, with accents such as US, UK and Indian English. On-screen text works in the same languages.
What resolution does it output?
On Higgsfield, Kling 4.0 generates at 720p or 1080p.
