Kling 4.0 launches with 30-second video

Kuaishou announced on 16 October 2026 that Kling 4.0, the new video-generation model of its Kling AI service, launched on 15 October. The release says it generates up to 30 seconds in a single pass, accepts up to 15 reference items in all (up to 10 images, 5 videos and 7 subjects) and up to 10 keyframes, takes prompts of up to 8,000 tokens and outputs up to 10-bit HDR. Video extension to two minutes, 4K and API access are announced as coming.

What the release says. Datelined Hong Kong, 16 October 2026 (the investor-relations page shows 17 October): Kling 4.0 "officially launched on October 15". The upgrades "center on three priorities: realism, control and professional quality". Steadier large-scale motion and complex camera movement; native generation of up to 30 seconds in a single pass; Omni Reference, "up to a total of 15 references, including up to 10 images, 5 videos and 7 subjects", combining images, videos, subjects and voices in one task, plus up to 10 keyframe inputs. Prompt input up to 8,000 tokens. On the audiovisual side the model "supports output of up to 10-bit HDR" and "will also support 4K, 1080p and the 21:9 ultrawide format". The video-extension feature "will soon support multiple extensions of up to two minutes". Kling 4.0 is integrated with Kling MCP and CLI so that AI agents can use Omni Reference and keyframe-to-video generation; its capabilities "will soon be made available to enterprise customers and developers through API access". The release adds that two works created with Kling AI won Gold Lions at the Cannes Lions festival in summer 2026, which, the release says, incorporated AI into its judging framework for the first time. What the record does not claim. The release gives no benchmark, names no resolution actually delivered at launch beyond the HDR bit depth, and gives no price or availability by country; the "breakthroughs" in steadiness of motion are the company's word. No sample video was viewed.