Kling Tests 30-Second AI Video Model With Select Users

Kling 4.0, the latest AI video-generation model from the Beijing-based Kling team, doubles the maximum clip length to 30 seconds and accepts more than a dozen types of direction, including text, images and existing video. Creators can set up to 10 keyframes to guide how a clip develops, with the tools aimed at maintaining visual consistency and controlling camera movement and scene continuity. The model is being tested with a select group; annual subscribers can use a lite version, with a broader rollout planned for October. Kling is preparing for a possible Hong Kong listing after raising $2.8 billion in its first funding round, while seeking to compete with ByteDance’s Seedance amid broader challenges monetizing AI video and pressure on parent Kuaishou’s business.