Ceramic Still Life
16:9Camera slowly orbits around the vase. Soft light shifts across the ceramic surface. The pampas grass sways gently. Shadows move elegantly. Smooth continuous motion, premium feel.
Kling 3.0
ID:
kling/kling-3.0Kling 3.0 video generation API by Kuaishou — native 4K at 60fps with multi-shot storyboarding, audio in 5 languages, and 3 quality modes.
Text to videoImage to video4KMulti-shot
Input
OutputView all
Output will appear here
Pricing
std
$0.09/s
pro
$0.12/s
4K
$0.44/s
Examples
Craftsman Portrait
16:9The craftsman slowly examines the bowl, turning it gently in his weathered hands. His eyes reflect years of wisdom. Subtle smile forms on his face. Dust particles drift in warm light. Breathing motion, blinking eyes.
Element Injection
1:1Stylized sunglasses resting on cracked desert ground under a dramatic sunset sky, reflective lenses catching the warm light
Overview
Kling 3.0 is Kuaishou's flagship video model, released in February 2026. It generates native 4K (2160p) video at 60fps — not upscaled from a lower resolution. The model supports multi-shot storyboarding with up to 6 camera cuts in a single generation, and produces audio in 5 languages. Three quality modes let you trade resolution for cost: std (720p), pro (1080p), and 4K (2160p). Clips range from 3 to 15 seconds, and up to 2 frame images can anchor the start and end of a sequence.
Use cases
High-resolution product videos and e-commerce hero content at native 4K. Multi-shot brand films with planned camera cuts and transitions. Character-driven narrative clips with multilingual audio. Image-to-video with first and last frame control for precise motion arcs.
Inputs
All parameters are passed in the input object of the run request.
| Parameter | Required | Description |
|---|---|---|
| prompt | Yes | Text description (1–2500 chars) |
| aspect_ratio | No | Default 16:9. Options: 16:9, 9:16, 1:1 |
| mode | No | Quality mode. Default pro. Options: std (720p), pro (1080p), 4K (2160p) |
| duration | No | Video length in seconds (3–15). Default 5 |
| image_urls | No | Up to 2 frame images (1 = first frame, 2 = first + last frame) |
Prompt tips
Structure multi-shot prompts as a shot list
Kling 3.0 handles up to 6 camera cuts. Write prompts as sequential shots: "Wide shot of the storefront. Cut to close-up of the sign. Pan across the interior." The model interprets these as distinct segments.
Use std mode for drafts, 4K for finals
Start with std (720p) during iteration — it's fast and cheap. Switch to 4K only for final renders. The pro mode at 1080p is a good middle ground for most production work.
Keep prompts under 2500 characters
The prompt limit is shorter than some competitors. Focus on essential scene details, camera directions, and key actions. Cut adjective-heavy descriptions that don't affect the visual output.
Limitations
- 4K mode significantly increases generation time and cost
- Only 3 aspect ratios supported (16:9, 9:16, 1:1)
- Prompt limit is 2500 characters — shorter than Seedance 2.0's 20000
- Multi-shot transitions may not always cut at the exact moments described
- 60fps output increases file size substantially at 4K resolution
FAQ
Is the 4K output truly native?
Yes. Kling 3.0 renders at 2160p natively — the frames are not upscaled from a lower resolution. This is reflected in the 4K quality mode option.
How do the two frame images work?
Provide one image to set the first frame. Provide two images to anchor both the first and last frames — the model generates video that transitions between them. Each image can be up to 10 MB.
What languages does the audio support?
Kling 3.0 generates audio in 5 languages — English, Chinese, Japanese, Korean, and Spanish. The language is inferred from the prompt context. Specify the language in your prompt if the model doesn't pick it up automatically.