Kling 4.0 adds 4K and 15-reference inputs
Kling's guide to its Kling 4.0 video model describes 4K output, stereo audio and up to 15 combined reference inputs, aimed at keeping characters and products consistent across shots.
Kling has published a guide to Kling 4.0, the next generation of its AI video model. The company says it improves motion, audio and reference-based control. According to the guide, the full model was scheduled for October 2026, and early access to a new creation experience began rolling out gradually on September 28.
Output and audio. Kling 4.0 supports 4K output and 21:9 ultrawide video. The company says 10-bit HDR at 1080p and 4K is "coming soon." It also claims higher-quality audio, two-channel stereo and more accurate lip-sync. The company says the model keeps subject structure and motion more stable during fast action and complex camera moves such as orbits and tracking shots.
Omni Reference. The expanded reference system takes up to 15 combined assets in one generation:
- up to 10 images, including character turnarounds, storyboards and wireframes
- up to 5 videos with a combined length of up to 30 seconds, including white-model and depth-map videos
- up to 7 subjects, of which up to 3 can be video-based, plus optional voice references
Kling says this keeps faces, clothing and product details consistent across changes in setting, lighting and framing. The same references can also be used to recreate trending video formats with a creator's own assets. In editing tasks, the full model accepts up to five video inputs, one of them primary, with a combined length of up to 30 seconds. Users can add, replace or remove elements and change backgrounds, colors or camera angles.
The guide is Kling's own promotional material. It includes no independent benchmarks, pricing or API details.