# Kling 4.0

> The newest tier — clips of three to thirty seconds at 720p or 1080p, from your words, from a frame, or from up to ten pictures as references.

**Kling 4.0** makes clips of three to thirty seconds at 720p or 1080p: from your words, from a first frame (with a middle and a last one if you add them), or from up to ten pictures it keeps as references. When Studio offers it, you find it under **Videos** on Studio's home and **Generation** in **Choose a model** — and, once the owner switches it on, in the chat too.

## What it takes and makes

### Make — words to video

| Setting | What it takes |
| --- | --- |
| Your words | Required — up to 10,000 characters |
| Length | 3 to 30 seconds — starts at 10 seconds |
| Quality | 720p or 1080p — starts at 720p |
| Shape | 21:9, 16:9, 9:16 or 1:1 — starts at 16:9 |
| Sound | Silent video |
| Makes | A video |

### Animate — a picture comes to life

| Setting | What it takes |
| --- | --- |
| Your words | Required — up to 10,000 characters |
| Frames | A first frame (required) and, if you like, a last frame |
| Length | 3 to 30 seconds — starts at 10 seconds |
| Quality | 720p or 1080p — starts at 720p |
| Shape | Follows your first frame |
| Sound | Silent video |
| Makes | A video |

### Reference — your pictures and clips guide a new video

| Setting | What it takes |
| --- | --- |
| Your words | Required — up to 10,000 characters |
| Pictures | Up to 10 pictures (Image 1, Image 2… in the order you add them) |
| Length | 3 to 30 seconds — starts at 10 seconds |
| Quality | 720p or 1080p — starts at 720p |
| Shape | 21:9, 16:9, 9:16 or 1:1 — starts at 16:9 |
| Naming your references | Type @ and pick one — one of each kind, read as “the picture”, “the clip” or “the sound” |
| Sound | Silent video |
| Makes | A video |

The price is never written here — prices change. It is on the model's page in Studio: on the **Make** button and beside each length, size or quality that changes it, before you press.

## Writing for it

- In **Make**, say the scene: the partner publishes no limit on your words, so they are sent whole — say what matters first anyway.
- In **Animate**, your frame decides the shape; the partner takes no shape of its own on that endpoint. Words are optional there, and a sentence about what moves helps.
- In **Reference**, point at each picture by its number — type `@` and pick — and say what each one gives the take: *"@Image2 is the jacket"*. A tag that names a picture you did not add binds nothing.
- There is no sound switch on these endpoints: a take with the model's own sound cannot be asked for, and the table above says so.
- A take with a reference *clip* is not offered yet: the partner bills it by a rule ROUTIX's price layer does not carry, so the door refuses it before any credit moves rather than guess.

Source: https://routix.studio/help/video-models/kling-4-0 · Updated 2026-10-30