Kling AI 4.0 Generates 30-Second Videos and Stereo Audio in One Go
Kling AI 4.0 is an AI video model that generates videos up to 30 seconds in a single pass and supports two-channel stereo audio with improved lip sync. It also accepts up to 15 reference assets and in
Kling AI 4.0 is a new-generation model focused on aligning the visuals and sound of short videos more closely. Kuaishou said the model can generate videos of up to 30 seconds in a single generation and supports high-quality two-channel stereo audio and improved lip-sync. The central change is the ability to review the coordination of voice, mouth movements, and performance together in dialogue scenes involving transitions or movement.
Using Reference Materials for Scene Design
Omni Reference combines up to 15 reference materials—including images, videos, people or objects, and voices—in a single task. It can use up to 10 images or up to five videos totaling 30 seconds. This makes it useful for maintaining a character’s appearance while determining camera movement or the pace of a scene, but having more reference materials does not automatically guarantee the intended result.
Flash as an Option for Draft Iteration
Kling 4.0 Flash is a variant that emphasizes speed and cost efficiency for frequent generation tasks. Flash suits the drafting stage, when multiple scene compositions and dialogue timings need to be compared quickly, while results requiring complex keyframes or precise integration of reference assets can be reviewed again with the standard model.
Expanded Video Length and Control Range
The officially released Kling AI 4.0 supports up to 10 keyframe inputs and prompts of up to 8,000 tokens. This is not simply an update that raises resolution; it reflects a shift in the production workflow toward designing scene flow through long single shots and multiple reference materials. However, when using multilingual dialogue and intonation, the naturalness of speech in each language and brand-specific phrasing still need to be checked separately.