Note: Kling O3 is Kuaishou's latest video generation model, building on the Kling family. Learn more at Kling AI and try it.
We've added Kling O3 I2V and Kling O3 Reference—two new video models for image-to-video with optional audio, end-frame control, and reference-based generation.
What's New
- Kling O3 I2V: Image-to-video from a start frame with optional end frame. Animate the transition with text-driven style and scene guidance. 3–15 second clips with optional synchronized audio.
- Kling O3 Reference: Transform reference images and elements into consistent video. Use @Element1, @Element2 for characters/objects and @Image1, @Image2 for style—with optional start and end frames. 3–15 seconds with optional audio.
- Regenerate Modal: The regeneration flow now supports optional last-frame images with preview tabs. Last-frame validation is applied per model—unsupported models ignore the end frame.
About Kling O3
Both models support:
- Start & End Frame Control: Provide a start frame (required) and optionally an end frame to guide the motion and composition.
- Optional Audio: Generate synchronized audio alongside the video.
- Flexible Duration: 3–15 seconds per clip.
- Reference Notation: In Kling O3 Reference, use @Element1, @Element2 and @Image1, @Image2 in your prompt to reference uploaded elements and style images.
How It Works
Kling O3 I2V: Select the model, add a start frame (and optionally an end frame), write your prompt, and generate. Ideal for animating a single image with controlled motion.
Kling O3 Reference: Select the model, add reference elements and images, use @Element1, @Image1, etc. in your prompt, optionally add start/end frames, and generate. Ideal for character-consistent or style-guided videos.
Learn more about Kling and Kuaishou's AI video models at kling.kuaishou.com.