Long-form
Up to 30 seconds per generation
- 3–30 seconds in one pass — a continuous take or a complete short story.
- Prompts up to 8,000 tokens, from one line to a full shot list.
Coming soon to Pika
Kling 4.0
Kling 4.0 by Kuaishou
Up to 30 seconds of video with stereo sound — from text, an image or up to 15 references — and edits to footage you already have.
First outputs at launch
Real Kling 4.0 generations land here the day it ships.
- Per generation, in one pass
- 30s
- Dialogue, music & effects, lip-synced
- Stereo
- Image, video & element references
- 15
- Keyframes to direct the shot
- 10
What it can do
Native audio
Sound and lip sync in the same pass
- Dialogue, music, effects and ambience as two-channel stereo.
- Lip sync stays locked to voices, dialogue and performances.
References
Up to 15 references in one video
- Up to 10 images, 5 videos and 7 elements — address each in the prompt.
- Up to 3 elements can come from video, and an element can carry its own voice.
Motion
Borrow the move from another clip
- A reference video guides performance, camera and pacing.
- Depth-map and white-model renders work as references too.
Direction
Pin the story beats with keyframes
- Up to 10 keyframe images, hit in order.
- Or pin just the first frame — or the first and last.
Editing
Edit a take, keep the rest of it
- Re-cast, re-set, remove an element or restyle — the scene holds.
- Up to 5 source clips, one of them primary — each 3–30s, 30s combined.
Speech
Dialogue in 10+ languages and accents
- English, Chinese, Japanese, Korean, Spanish, Portuguese, German, French, Indonesian, Hindi and more — plus accents and dialects like British English and Cantonese.
- On-screen text stays legible as the camera moves.
Camera
Push-ins, orbits and tracking shots
- Fast action stays stable through tracking and orbit moves.
- 21:9, 16:9, 1:1 or 9:16 — or let the model choose.
Inputs in, video out
Input · 2 images
@Image1
@Image2
Starter prompt
@Image1 is the character sheet and @Image2 is the storyboard. Follow the storyboard panel by panel, keeping the character on model in every shot.
Result
Character sheet to short
Ready on day one
Kling 4.0 opens in Create Video the day it ships. Start on Kling 3.0 today and switch models in one click.
- Pick a mode
Text to video, image to video (first frame, or first and last), reference to video, or edit a clip you already have.
- Write it like a shot brief
Describe action, camera and sound. Address references as @Image1, @Video1, @Element1. Set 3–30s and an aspect ratio.
- Draft on Flash, finish on 4.0
Iterate quickly at 720p, then render the keeper at 1080p with stereo audio already in sync.
Which Kling to use
| Model | Best for | Length | Resolution | Audio | On Pika |
|---|---|---|---|---|---|
| Kling 4.0 | Final renders, long takes, references, editing | 30s | Up to 4K | Stereo, lip sync | Coming soon |
| Kling 4.0 Flash | Fast, frequent drafts | 20s | 720p | Native | Coming soon |
| Kling 3.0 | Multi-shot storyboards | 15s | Up to 4K | Native | Live today |
Questions
As soon as Kling opens access. We email everyone who signed up the day it goes live.
Pricing is published at launch. Kling 4.0 Flash is the lower-cost option for drafts.
Kling 4.0 generates 3–30 seconds per pass; Flash generates 3–20. Kling has announced extension up to 2 minutes as coming later.
720p, 1080p or 4K, in 21:9, 16:9, 1:1 or 9:16. Flash outputs 720p. Kling has announced 10-bit HDR at 1080p and 4K as coming later.
Double the length (30s vs 15s), stereo audio, 21:9, up to 15 references and video editing. Kling 3.0 stays on Pika.
Be first to try it
Try Kling 3.0Last updated October 20, 2026 · Model details from Kling AI’s Kling 4.0 release note.