Kling 4.0 adds longer clips and more production controls

Kling 4.0 supports up to 10 keyframe inputs for defining key moments, character states, scene changes and story beats.[3] It also accepts up to 15 combined reference assets, including images, videos, elements and voice styles.[3] Kling AI says users can make targeted edits to expressions, movement, camera choices, style and backgrounds while keeping the rest of a scene stable.[1] TechNode describes the distinction as references guiding how a video looks and feels, while keyframes define how its story unfolds.[5] This gives teams a more explicit way to specify a shot, but not proof that every complex scene will follow instructions reliably.[5]

A single Kling 4.0 generation can produce a video up to 30 seconds long, double Kling 3.0’s stated 15-second maximum.[2] Kling AI also says users can extend videos multiple times up to two minutes, but marks that feature as coming soon.[1] The shorter Kling 4.0 Flash option has a different trade-off: FrameSignal reported that it makes clips up to 20 seconds at lower picture quality and does not support still-image guidance.[4] Teams evaluating it should distinguish the full model’s controls from Flash’s availability and limitations rather than assuming the two versions offer the same workflow.

Kling AI says Kling 4.0 entered early access, with a wider rollout planned for October.[1] Kling 4.0 Flash was due to become available to Ultra Yearly subscribers starting September 28.[1] FrameSignal reported that Kling had not disclosed the full model’s price or exactly when in October everyone could use it.[4] Kling AI lists support for two-channel stereo audio, more accurate lip-sync and multiple languages, and says the model can render multilingual text, emojis and logos through camera movement and scene changes.[1] 10-bit HDR at 4K and 1080p is marked as coming soon.[1]

Kling AI says the model can use a reference video’s pacing, shot composition and visual expression while replacing products, characters and other assets.[1] That may suit teams exploring variations on an existing visual direction, but it does not establish how consistently the feature performs across production work. FrameSignal noted that the examples it had seen were all made and selected by Kling, leaving everyday performance uncertain.[4] TechNode also cautioned that the value of references and keyframes in reducing repeated generations depends on scene complexity.[5] Its reporting says human judgment remains necessary for story development, shot selection and final editing.[5]