Run Kling 4.0 in your browser: up to 30 seconds in one take, with dialogue and sound built in.
A 5-second 720p clip is 143 credits; the $19.90 pack (600 credits) covers four.
Six clips from Kling's own 4.0 release and two we made with the generator above. Click Use template on a clip to load its prompt into the generator.
Kling 4.0, often searched as Kling 4, is the newest video model from Kuaishou's Kling AI. It renders up to 30 seconds in one pass with dialogue, ambience and effects generated alongside the picture, follows long prompts closely, and can be steered with start and end frames or a set of reference images and videos. On ImagesArt it works as a Kling AI video generator without a Kling subscription: you pay per second with credits and can switch to Kling 3.0, Seedance or Veo in the same account.
Source: Kling AI release notes
Twice the length. Kling 3.0 stops at 15 seconds per generation (12 on ImagesArt). The new model runs to 30, so a scene can set up, play out and land in one take.
Sound in every clip. Speech, ambience and effects come out of the same generation. Put a line in quotes and the character says it with matching mouth movement.
More ways to direct it. Start and end frames, up to 9 reference images and up to 3 reference videos on ImagesArt. The model also takes up to 10 keyframes; ImagesArt supports the first and last frame for now.
Exactly what the generator at the top of this page offers.
Set any length from 3 to 30 seconds and get one continuous generation, not clips stitched together. This 28-second dragon attack from Kling's release keeps the same creature, city and fire from the first frame to the last.
Every clip comes with sound generated in the same pass. Write the line in quotes and the character speaks it with matching lip movement, over room tone and effects that fit the scene. That makes Kling 4.0 text to video a one-step job: no separate voice-over or sound design.
Upload where the shot begins and where it ends, and the model fills in everything between them, sound included. It's the most controllable way to run Kling 4.0 image to video.
Add up to 9 images of people, products or places and up to 3 reference videos, then call them Image 1, Image 2 or Video 1 in the prompt. Here one character photo and a storyboard became a three-shot drink ad.
Choose Text/Image to Video, Start-End Frame or Reference. Add a photo, two frames or up to 9 reference images, or just write the scene.
Say what happens and when, how the camera moves and what we hear. Put spoken lines in quotes so the character lip-syncs them.
Pick 3 to 30 seconds and 720p or 1080p, check the credit cost on the Create button, and the MP4 lands in your history.
The model follows long, specific prompts closely. Build yours from these four parts.
Start with who or what is on screen and what they do. Use plain verbs — runs, turns, lifts the cup — rather than adjectives about the mood.
For clips over 10 seconds, write a short shot list: "0–5s: …, 5–12s: …". Give each range one main change so nothing feels rushed.
Name the move and where it ends: slow push-in, tracking shot, crane up, handheld. One move per range keeps the shot readable.
Put every spoken line in quotes and say who speaks it. Add ambience and effects — rain, footsteps, crowd chatter — because the model always generates sound.
Each one opens in the generator above with the mode, aspect ratio and length already set. Swap in your own subject.
0–8s: a young woman waits alone at a rainy bus stop at night, checking her phone, neon reflections on the wet street. 8–16s: a man runs up holding a broken umbrella and stops a few steps away; they look at each other. 16–24s: close-up as he says: "I got your message. Did you mean it?" She hesitates, then nods. 24–30s: the bus pulls in, its headlights wash over them, and she steps toward him instead. Rain, distant traffic, soft piano.
A vertical selfie video of a young man in his kitchen holding a new pair of wireless earbuds. He opens the case toward the camera and says: "Six hours on one charge, and they actually stay in when I run." Natural kitchen light, slight handheld shake, quiet room tone.
0–6s: a wide drone shot glides over a misty pine forest at sunrise. 6–11s: the camera descends toward a wooden cabin by a lake, smoke rising from the chimney. 11–15s: a slow push-in on the porch as the door opens. Birdsong, wind in the trees, a low cello note.
Keep the person, outfit and setting from the photo. 0–5s: they look up toward the camera and smile, a gentle push-in. 5–10s: a breeze moves their hair and the light warms slightly as they say softly: "Hey, you made it." Quiet room tone.
"The girl" is ambiguous when you upload four photos. Call them Image 1, Image 2 or Video 1, numbered in the order you added them, so the model knows which file is which.
Packing three actions and two camera moves into a short clip makes the motion jumpy. Give the shot more seconds, or split it into more time ranges.
The model always adds audio. If you don't describe it you get generic background noise, so name the ambience you want, or ask for "no music" if you plan to add your own.
All three run on ImagesArt with native audio and bill per second. They differ in clip length, top resolution, reference inputs and price.
| Kling 4.0 | Kling 3.0 | Seedance 2.5 | |
|---|---|---|---|
| Longest clip | 30 s | 12 s | 30 s |
| Top resolution | 1080p | 4K | 1080p |
| Credits per second at 720p | 28.5 | 18 | 63 |
| Credits per second at 1080p | 42.5 | 23 | 158 |
| 5 seconds at 720p | 143 credits | 90 credits | 315 credits |
| Reference inputs | Up to 9 images and 3 videos | Up to 7 images | Up to 9 images, 3 videos and 3 audio clips |
| Sound | Always on | Always on | On by default, can be switched off |
All three take text, a photo, first and last frames and reference images. Kling 3.0 also renders 4K at 67 credits per second; Seedance 2.5 also renders 480p at 28 credits per second.
The shot runs past 12 seconds and you want sound at under half the per-second price of Seedance 2.5
The clip fits in 12 seconds, you want 4K, or you want the lowest price per second
You want 480p drafts, reference audio or dialogue in languages other than English
Billed per second of video; the rate depends only on resolution, and sound is always included. Here are the common lengths.
| Length | 720p | 1080p |
|---|---|---|
| Per second | 28.5 | 42.5 |
| 5 seconds | 143 | 213 |
| 10 seconds | 285 | 425 |
| 15 seconds | 428 | 638 |
| 20 seconds | 570 | 850 |
| 30 seconds | 855 | 1,275 |
Credits = seconds × the rate for your resolution: 28.5 at 720p and 42.5 at 1080p, rounded to the nearest credit. Audio is always included, and the Create button shows the exact cost before you commit.
Adding reference videos switches to 17 credits per second at 720p (25.5 at 1080p), counted on the longer of twice the output length or video plus output. A 5-second clip with a 4-second reference video bills 10 seconds: 170 credits at 720p.
If a render fails, nothing is deducted. You only pay for videos that land in your history.
Buy a one-time credit pack or subscribe for monthly credits. This model starts at 720p, which comes with a Starter plan or any credit pack; 1080p comes with Pro or any credit pack.
Credits
Features
Credits
Features
Credits
Features
Credits
Features
Thirty seconds with sound covers formats that used to need several clips and an editor.
Stage a full beat — setup, line, reaction — in one 20 to 30 second take, with the dialogue already voiced.
Give the Reference tab your product photo and a storyboard to get a multi-shot spot that keeps the packaging and logo readable.
Long, continuous camera moves through a scene give editors full takes to cut to the beat.
Generate 9:16 talking clips and skits for Reels, Shorts and TikTok with the sound already in, ready to post.
Bring a portrait, product shot or travel photo to life from the first frame, or set the last frame too to control where it ends.
Try out camera moves, blocking and pacing for a shoot before you book the crew.
Use it next to Seedance 2.5, Veo 3.1, Grok Imagine and Wan 3.0 on one page with one credit balance.
Open the AI image to video generatorUp to 30 seconds in one take, native audio and lip-sync, 720p or 1080p — right in your browser, no software to install.
Compare plans and credit packs