Scenes with more than one person
Dialogue between 2 characters, a presenter and a product, a crowd that has to hold together.
Create with Kling 3.0
Latest
Kling on Magnific
Kling 3.0: cinematic realism with native audio
Kling 3.0 is Kuaishou's current flagship video model, generating video with synchronized native audio from a prompt or an image, with references for up to 3 characters in the same scene.
- 4/5Trustpilot · 3,100+ reviews
- ISO 27001Certified · GDPR compliant
- 100% yoursCommercial use included
Made with Kling 3.0 on Magnific
What's new
What's new vs Kling 2.6
Improved realism and physics. Character references for up to 3 people in one scene. Multi-shot clips of up to 6 shots and 15s, at up to 4K. And better native audio than 2.6's first implementation.
See the full Kling changelogSpecs
Kling 3.0 specs
The same 9 lines on every version page, so 2 versions read side by side. A dash means the value isn't confirmed on Magnific yet.
| Inputs | Text, ImageStart frame, end frame, and up to 3 references |
|---|---|
| Max length | 15sUp to 6 shots in one clip |
| Resolution | 720p, 1080p, 4K |
| Frame rate | — |
| Native audio | Yes |
| References | Up to 3 (character, product, or image) |
| Aspect ratios | 16:9, 9:16, 1:1 |
| Credits / s | 70 / 90Standard / Pro. With native audio: 105 / 140. 4K: 400 |
| Plans | Credits on any paid plan |
Credits per second at list price. Where a line shows 2 figures, the note under it says what each one is.
Compare every Kling versionBest for
Best for scenes with more than one person
The multi-person references are the reason to pick this over anything older. The audio and the physics are why you stay.
Anything with speech
Audio comes out of the same generation. No separate voice track, no manual sync.
Generate a talking scene
Product in motion
Where the physics have to look right rather than approximately right.
Put a product in motionNumbers
How Kling 3.0 actually gets used here
Measured on Magnific over 30 days, across the web, the apps, the API, MCP, and agents.
01
69,000
The Kling version most people use
69,000 people generated with 3.0 in 30 days, more than with any other version, 2.5 included.
02
1 in 5
Clips that ship with native audio
19% of 3.0 generations come out with sound in the same pass: voice, effects, and ambience. Audio isn't free: it adds about 50 credits a second at either quality.
03
95%
Start from an image
95% of 3.0 clips begin with a start frame. The typical clip is 5s.
Source: Magnific usage data, September 5 to October 4, 2026. Shares are of generations unless a figure says seconds.
Kling 3.0, answered
When is 3.0 worth the credits over 2.5?
When you need audio, more than one person in frame, or references at 1080p and 4K. Otherwise 2.5 gives you a longer clip for less.
What is Kling 3.0?
Kuaishou's current flagship video model. It generates video with synchronized native audio from a prompt or an image, takes up to 3 character references, and runs clips of up to 6 shots and 15s at up to 4K.
How much does Kling 3.0 cost?
70 credits a second at Standard and 90 at Pro. Native audio adds about 50 a second: 105 and 140. 4K is 400 a second. A 5s clip at Pro 1080p without audio is 450 credits.
What inputs does Kling 3.0 take?
A text prompt, a start frame, an optional end frame, and up to 3 character, product, or image references. It doesn't take video as input: that's Kling 3.0 Omni.
Which plans include Kling 3.0?
Every paid plan, on credits. There's no Kling subscription and no separate account: pick the version in the video generator and go.
Your next scene, with its own sound
Kling 3.0, in the same generator as every other version.