Kling 4.0: 30-second AI video you can direct

Built on Kuaishou's Kling 4.0. Place a start frame, an end frame and up to 8 middle frames along a 30-second timeline, bring in up to 15 image and video references, and render in 1080p, including 21:9.

What is Kling 4.0?

Kling 4.0 is Kuaishou's latest AI video model, available in the Artlist AI Toolkit. It generates up to 30 seconds of 1080p video in a single pass, with a start frame, an end frame and up to 8 middle frames to steer the shot, up to 15 image and video references, and stereo audio with dialogue in 9 languages.

Start a 30s video
Kling 4.0 still: woman in a white dress on an alpine meadow under cumulus clouds

What you can make with Kling 4.0

Kling 3.0 gave you 15 seconds and a start and end frame. Kling 4.0 gives you a half-minute scene and control over what happens in between.

  • A beginning, middle and end

    Write a 30-second scene as timed beats and get one continuous take, not three clips stitched together.

  • Decide what happens at second 12

    Place up to 8 middle frames anywhere on the timeline, between your start and end frame, to time a reveal, a camera move or a shot change exactly where you want it.

  • Keep the same face and product

    Load your character and your product once, and they hold across the whole take.

  • Action that holds together

    Fights, sports, dance and crowd scenes stay readable instead of dissolving mid-move.

  • Dialogue your audience actually hears

    Stereo sound and lip-synced speech in 9 languages, including regional accents, with no second pass for audio.

Kling 4.0 vs Kling 3.0, Seedance 2.5 and Veo 3.1

Kling 4.0
Kling 3.0
Seedance 2.5
Veo 3.1
Developer
Kuaishou
Kuaishou
ByteDance
Google DeepMind
Max length, one pass
30 s
15 s
30 s
8 s (+7 s per extend)
Max resolution
1080p
Up to 4K
1080p
1080p (4K on Fast)
Keyframes
Up to 10: start, end and up to 8 middle frames
Start and end frame
First and last frame
First and last frame
References per generation
15: 10 images + 5 videos
Start frame + reference images
50: 30 images + 10 videos + 10 audio
Ingredients: character, object, style
Native audio
Stereo, same pass
Yes, same pass
Yes, same pass
Yes, same pass
Dialogue languages
9
5
11
Not specified
Aspect ratios
21:9, 16:9, 9:16, 1:1
16:9, 9:16, 1:1
21:9, 16:9, 9:16, 1:1
16:9, 9:16
Pick it when
You need to direct a long take shot by shot
You want the proven Kling look at 15 s
You have many references or need 11 languages
You want Google's look and 4K on short clips

Kling 4.0 capabilities

Everything below is available in Kling 4.0 on Artlist.

Place your keyframes

  • 30-second single take

    Set any length from 3 to 30 seconds in one generation, double Kling 3.0's limit. Long enough for a product demo, a trailer beat or a full social spot.

  • Image-to-video with middle frames

    Start from an image, add an end frame and place up to 8 middle frames anywhere on the timeline. Each frame anchors a camera move, a shot change or a reveal.

  • Omni: up to 15 references

    Combine up to 10 images and 5 video clips in one generation. Video references can run 3 to 30 seconds each, 30 seconds total.

  • Complex motion

    More stable than Kling 3.0 in fights, sports, dance and large-scale action, where limbs, props and crowds tend to break.

  • Stereo audio in the same pass

    Two-channel sound is generated with the picture, with tighter lip-sync than Kling 3.0. Switch it off when you plan to score the clip yourself.

  • Dialogue in 9 languages

    English, Chinese, Japanese, Korean, Spanish, Portuguese, German and French, with accents and dialects such as Cantonese and Indian English. Put lines in quotes and name the language.

  • On-screen text and emoji

    Renders readable text in the same languages, plus emoji, for titles, captions and signage inside the frame.

  • 21:9 widescreen, up to 1080p

    Choose 21:9, 16:9, 9:16 or 1:1 at 720p or 1080p. 21:9 is new to Kling and suits trailers and cinematic cutdowns.

Frequently asked questions

Kling 4.0 is Kuaishou's AI video model, the follow-up to Kling 3.0. It generates up to 30 seconds of 1080p video in one pass, with a start frame, an end frame and up to 8 middle frames, up to 15 image and video references, and stereo audio with dialogue in 9 languages. On Artlist it runs inside the AI Toolkit alongside Seedance 2.5 and Veo 3.1.

Kling 4.0 doubles the clip length from 15 to 30 seconds and builds on start-and-end-frame control with up to 8 middle frames anywhere on the timeline. It also adds stereo audio, a 21:9 aspect ratio, 9 dialogue languages instead of 5, and up to 15 references in one generation. Motion in action scenes is more stable.

A single Kling 4.0 generation runs from 3 to 30 seconds. For a long clip, write the prompt as timed beats, for example 0-8 seconds wide establishing shot, 8-15 seconds slow push to close-up, so the model knows what each stretch of the take should do.

Yes. Kling 4.0 generates stereo audio in the same pass as the video, including dialogue, ambience and effects, with tighter lip-sync than Kling 3.0. Dialogue works in 9 languages, with accents and dialects such as Cantonese and Indian English. Audio is optional, so you can switch it off and score the clip with Artlist music instead.

Not yet. Kling 4.0 outputs at 720p and 1080p. Kuaishou has announced 4K and 10-bit HDR output, but neither is available at launch. For 4K today, Kling 3.0, Seedance 2.0 and Veo 3.1 Fast in the Artlist AI Toolkit are the options.

Open the Artlist AI Toolkit, choose video, and select Kling 4.0 from the model list. Write a prompt, add a start image, middle frames or Omni references, set the length and aspect ratio, and generate. Finished clips can be scored with licensed Artlist music and SFX in the same workspace.

Still have questions? We're here to help.