Kling 4.0

Kling 4.0 is the new generation of Kuaishou’s video model: a cinematic image, precise prompt adherence and native understanding of video references. Clips from 1 to 15 seconds in three frame formats, a first and a last frame, a reproducibility seed. Generation is asynchronous: POST /v1/videos returns a job, then you poll the status and download the mp4.

Parameters

Field	Description
`model`	`kuaishou/kling-v4.0` (the short `kling-4` / `kling-4.0`)
`prompt`	Scene description, required; up to 2500 characters
`seconds`	An integer from 1 to 15; defaults to 1
`size`	One of the six sizes below; defaults to `1280x720`
`input_reference`	Image-to-video: one image as the starting frame, `{"image_url": "data:image/...;base64,..."}`
`input_references`	Reference-to-video: 1 to 9 images, `data:` URLs
`input_video`	Video reference: `{"video_url": "https://..."}` or a `data:` URL — motion is carried over to the character in your images
`input_audio`	Audio reference: `{"audio_url": "https://..."}` or a `data:` URL — the soundtrack follows the track’s rhythm and mood
`seed`	Reproducibility seed: an integer 0…2147483647

Resolutions and frame formats:

Aspect	720p	1080p
`16:9`	`1280x720`	`1920x1080`
`9:16`	`720x1280`	`1080x1920`
`21:9`	`1680x720`	`2520x1080`

Examples

import requests

job = requests.post(
    "https://api.mixen.ai/v1/videos",
    headers={"Authorization": f"Bearer {API_KEY}"},
    json={
        "model": "kling-4",
        "prompt": "A drone circles a lighthouse on a rocky shore at sunrise, light mist over the water",
        "seconds": "10",
        "size": "1920x1080",
    },
).json()
print(job["id"], job["status"])

curl https://api.mixen.ai/v1/videos \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kling-4",
    "prompt": "A drone circles a lighthouse on a rocky shore at sunrise, light mist over the water",
    "seconds": "10",
    "size": "1920x1080"
  }'

Polling and download

curl https://api.mixen.ai/v1/videos/video_... \
  -H "Authorization: Bearer $MIXEN_API_KEY"

curl -L https://api.mixen.ai/v1/videos/video_.../content \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -o kling4.mp4

Statuses: queued → in_progress → completed or failed. A sensible polling interval is 5–10 seconds. For the full cycle see the video guide.

Editing and extending a clip (Edit / Extend)

The kuaishou/kling-v4.0-edit and kuaishou/kling-v4.0-extend tiers work only with an input clip (input_video, no images):

Edit — re-shoots your clip from the prompt. seconds = the output duration; resolution is chosen as usual.
Extend — extends the clip: seconds = how many seconds to add; the output is the source stitched with the continuation.

{
  "model": "kuaishou/kling-v4.0-edit",
  "prompt": "Add golden sunset light, warm cinematic grade",
  "seconds": "5",
  "size": "1280x720",
  "input_video": {"video_url": "https://example.com/clip.mp4"}
}

{
  "model": "kuaishou/kling-v4.0-extend",
  "prompt": "Continue the scene naturally",
  "seconds": "5",
  "size": "1280x720",
  "input_video": {"video_url": "https://example.com/clip.mp4"}
}

Poll and download as usual. The clip is up to 20 MB; the dashboard offers the same modes on Kling 4.0 (Text / Frames / References / Edit / Extend), with Extend’s duration meaning the seconds to add.

Pricing

Billed per second, the rate depends on the resolution: 720p — 18.85 ₽ per second, 1080p — 28.21 ₽. The web dashboard and bot accept several references per request (up to 9 images, 3 clips and 3 audio tracks); the API takes one of each; billing follows the actual cost of each input. The Edit/Extend tiers have their own per-second rates (720p/1080p only): Edit ≈ 30.2 / 45.3 ₽ per second, Extend ≈ 27.7 / 41.5 ₽. Current prices — the catalog.

Strengths and limitations

First and last frame: the clip starts and ends on the given images — available in the web dashboard and bot.
Video and audio references: motion from a clip and rhythm from a track are carried over to the character in your images.
A continuous 1–15 second duration and three frame formats, including the wide 21:9.
A reproducibility seed for repeatable takes.
The prompt is capped at 2500 characters — long scripts are better kept short.
No resolutions above 1080p.

All models — in the catalog. The general video workflow — the video guide.