Mixer AI
Kling 4.0
A clip of up to 30 seconds from text and references or from a chain of keyframes: named characters, voice samples and exact frames.
Updated
When to choose it
Kling 4.0 works in two modes in Mixer AI. Text and references builds a scene from a prompt, photos, videos, characters and voices: each hero gets a name and the prompt refers to it as @Name. Keyframes leads the clip from a first frame to a last frame through middle frames. Pick the mode first: it decides which materials you need and how the price is counted.
- Characters by name: up to 7 heroes or objects, each with 2–4 photos or one video clip; the prompt calls them as @Name.
- Up to 30 seconds at 720p or 1080p is enough for a scene with several actions.
- Keyframes: up to 8 middle frames between the first and the last frame set the route of the motion.
Modes and settings
- Text and references: up to 10 reference photos and 5 videos, up to 7 characters (2–4 photos or one 3–8 second video clip each) and up to 3 voice samples. The Video sound switch applies only to reference videos.
- Keyframes: a first frame, a last frame and up to 8 middle frames. The total number of frames depends on the duration and never exceeds 10.
- Duration 3–30 seconds, 720p or 1080p, aspect ratios Auto, 16:9, 9:16, 1:1 and 21:9; you can request 1, 2 or 4 results at a time.
- Billing is per second, with a 3-second minimum. A reference video lowers the rate, but you pay for the result seconds plus the larger of two values: the input video length or the result length.
Price in coins
| Version | Quality / size | Input | Coins / second |
|---|---|---|---|
| Reference | 720p | Text and references | ≈ 20.09 |
| Reference | 720p | With video reference | ≈ 12.76 |
| Reference | 1080p | Text and references | ≈ 28.7 |
| Reference | 1080p | With video reference | ≈ 18.22 |
| Keyframes | 720p | Keyframes | ≈ 20.09 |
| Keyframes | 1080p | Keyframes | ≈ 28.7 |
Billing is per second, not per fixed-length clip. The rate is multiplied by the billable duration; the complete request total is rounded up to a whole coin once.
With a reference video you pay for the result seconds plus the larger of two values: the input video length or the result length, at the With video reference rate. The minimum is 3 seconds.
≈ marks a fractional rate rounded to two decimal places for display; billing uses full precision.
In US dollars: from ≈ $0.221 per second. 1 coin ≈ $0.011.
What you can make
- A clip with a recurring hero: upload the hero's photos as a character and refer to it in different parts of one prompt.
- Animation from a storyboard: first, last and middle frames set the path and the prompt describes the motion between them.
How to direct the result
- Give characters short, different names and refer to them in the prompt as @Name, with the same letter case as in the form.
- In Keyframes mode describe only the motion between frames: the look is already set by the images.
Example: two heroes in one scene
@Mila walks up to a cafe table where @Oleg sits and puts an envelope in front of him. Static camera, medium shot. Oleg looks up, takes the envelope and smiles. Clothes and faces of the heroes stay the same. Daylight from the window.
What to keep in mind
- The modes do not mix: Keyframes accepts no video, and characters and voices belong only to Text and references mode.
- A hero photo must be at least 300 pixels on the short side; a video character needs at least 700 pixels and one continuous shot.
Sources
Model facts verified: 2026-10-24.
Guides and comparisons using this model
FAQ
How is Text and references different from Keyframes?
The first mode builds a scene from a prompt, photos, videos, characters and voices. The second builds the clip from frames: first, last and middle ones. Videos and characters exist only in the first mode.
How many characters can I use?
Up to 7. Each needs 2–4 photos or one video clip of 3–8 seconds. You can add up to 3 voice samples.
How do I refer to a character in the prompt?
Write @ and the name set in the form, for example @Mila. Letter case matters; use short, different names.
How many seconds are billed with a reference video?
The result seconds plus the larger of two values: the input video length or the result length. The calculation uses the With video reference row of the table.
What is the minimum duration?
3 seconds. It is also the minimum billable time.
What are the requirements for character photos and videos?
A photo needs at least 300 pixels on the short side; a video clip needs at least 700 pixels and one continuous shot without cuts.
What changes the price?
The coin rate per second depends on quality, mode and, for some models, sound. Multiply it by the billable duration. Where input video is charged, its duration also counts. The complete request total is rounded once.
Do I need a separate model subscription?
No. In Mixer AI, you top up one coin balance and use it for the tasks you need. You do not need a separate subscription to this model.