Kuaishou Kling's new-generation video generation model, focused on realism, controllability, and professionalism. It excels at large-scale motion and complex camera work, keeping footage stable and coherent in high-speed motion, tracking, and orbiting shots. It natively generates 3–30 seconds per run and supports long takes and continuous narrative. Keyframe mode supports the first frame, first and last frames, and intermediate frames at any point on the timeline, up to 10 frames. Omni Reference can combine up to 15 assets per task (images, videos, subjects, and voice tones) to precisely reuse appearance, motion, camera movement, and rhythm, and supports local edits to facial expressions, motion, background, and more in existing videos. Audio and video are generated together with dual-channel stereo and multilingual lip sync; prompts can be up to 8,000 tokens. Output is 720p/1080p, supports a 21:9 ultra-wide aspect ratio, and is suited for advertising and marketing, film and TV shorts, short dramas, and e-commerce content creation.
Resolution
720p
1080p
Without video input
35% off¥0.65/ sMarket: ¥1
¥0.975/ sMarket: ¥1.5
With video input
35% off¥0.39/ sMarket: ¥0.6
¥0.585/ sMarket: ¥0.9
Input
Text / Image / Video / Audio
Output
Video
生成视频的时长（秒），本模型可选 3-15，默认 5
生成视频的分辨率，本模型可选 720P、1080P，默认 720P
生成视频宽高比，文生视频可选 16:9、9:16、1:1，默认 16:9
是否生成音频，默认关闭；3.0 系列 720P 起支持有声
模型随机种子
是否自动优化提示词，开启后自动优化 prompt 以提升生成质量
点播输出配置；子字段名必须与上游请求字段一致（运行时按对象整体透传，不会重命名子键）
Optional. Slots follow the model's input_references declaration; upload a local file or paste a public URL.
Last frame requires a first frame URL
Multiple allowed; referenced as Image 1, Image 2... in the prompt
Kling
The VIDEO 2.6 model generates visuals, natural voiceover, matching sound effects, and ambient atmosphere in a single pass, connecting the world of sound with the world of vision. Whether you type text or upload an image, it instantly creates a complete dynamic video with sound, rhythm, and immersion.
| Resolution | Output price |
|---|---|
| 720p | ¥0.1935/ s |
| 1080p | ¥0.3226/ s |
Kling
Kling Video 3.0 features a revolutionary intelligent storyboard system that automatically orchestrates multi-shot sequences to deliver cinematic transitions. It generates videos up to 15 seconds in a single pass, with enhanced character consistency, native-level text overlay, and multilingual audio mixing.
| Resolution | Without video input | With video input |
|---|---|---|
| 720p | ¥0.3871/ s | ¥0.5806/ s |
| 1080p | ¥0.5161/ s | ¥0.7742/ s |
| 4k | ¥1.9355/ s | ¥2.9032/ s |
| 2k | ¥0.6452/ s | ¥0.9677/ s |
Kling
Kling-3.0-Omni is the flagship model, offering exceptional video quality, advanced artifact understanding, and enhanced artifact correction. Built on Video 3.0, Omni delivers ultimate quality for professional creators, with finer details, smoother motion dynamics, and a higher success rate.
| Resolution | Without video input | With video input |
|---|---|---|
| 720p | ¥0.3871/ s | ¥0.5806/ s |
| 1080p | ¥0.5161/ s | ¥0.7742/ s |
Kling
Kling 4.0 Flash is the lightweight video generation model in Kuaishou's Kling 4.0 series, built for high-frequency creative work with a focus on generation speed and cost effectiveness. It supports text-to-video, image-to-video (first frame) and omnipotent reference-to-video: a single task can combine up to 15 assets (up to 10 images, 5 videos and subjects), reliably reusing characters, product appearance and motion/camera work, and it can also make local edits to the subject, background and style of existing videos; prompts can be up to 26,000 English characters. Each generation runs 3–20 seconds, outputs 720p, supports 21:9 / 16:9 / 1:1 / 9:16 and adaptive aspect ratios, and produces audio and video together. It suits high-volume social media short video production, e-commerce asset creation and rapid creative validation.
| Resolution | Without video input | With video input |
|---|---|---|
| 720p | ¥0.26/ s | ¥0.39/ s |