Kling AI released its Kling 4.0 video model, The Decoder reports. It generates single videos up to 30 seconds, extends prompt understanding to 8,000 tokens, and adds two-channel stereo audio and lip sync in more than ten languages. A lighter Flash variant is also out.