Kling AI releases Kling 4.0 with stereo audio, multilingual lip sync and a Flash variant
AI video generator Kling AI has released version 4.0, which the company says brings higher resolution, more precise controls and a fast Flash variant that should close the gap between AI-generated and professional video.
On the generation side, users can now make up to 30 seconds of video in a single pass. The system accepts up to ten keyframe images as input to pin down scene transitions, character states and story beats, and prompt comprehension has grown to 8,000 tokens.
The upgraded Omni Reference system takes up to 15 reference assets in a single generation, according to Kling AI. That covers up to 10 images, 5 videos with a combined maximum of 30 seconds, and 7 elements. Storyboards, wireframes, image grids, white-model videos and depth-map videos all work as references. A new Creative Recreation feature lets users take existing videos as templates and swap out products or characters; the company says individual elements such as facial expressions, body movements, camera angles or backgrounds can be tweaked without affecting the rest of the clip.
Audio and speech are the other big part of the update. Kling 4.0 adds two-channel stereo audio and improved lip sync across more than ten languages, including German, English, Japanese, Korean and Spanish, along with various dialects and accents, according to the company. In-video text generation now works in multiple languages and stays stable during camera movements.
Alongside the main model comes Kling 4.0 Flash, designed for faster generation at lower cost. The trade-off is a ceiling of 720p and 20 seconds of video. The creation interface has been redesigned, and Kling 4.0 can now be reached through MCP and CLI, which makes it possible to build automated workflows with AI agents.
Some announced features are not live yet. 10-bit HDR output at 4K and 1080p, and a video extension function that would stretch clips to up to two minutes, are both listed as coming soon by Kling AI. Pricing for the Kling 4.0 API has not been announced, and API support is also listed as coming soon.
The Decoder compared Kling 4.0 with Wan 3.0 and Seedance 2.5, both from Chinese competitors, in a video. Its verdict: Kling 4.0 holds its own but does not pull ahead, especially in logical shot sequencing, physics and natural-sounding dialogue. The outlet adds that this could be an early sign that the fast pace of progress in AI video models is starting to level off, though the next Wan and Seedance releases will tell more.
Key facts
- Kling 4.0 generates up to 30 seconds of video in one pass, accepts up to ten keyframe images and understands prompts of up to 8,000 tokens.
- Omni Reference takes up to 15 reference assets per generation (up to 10 images, 5 videos totalling at most 30 seconds, and 7 elements), per Kling AI.
- The model adds two-channel stereo audio and improved lip sync across more than ten languages, according to the company.
- Kling 4.0 Flash is built for faster, cheaper generation but is capped at 720p and 20 seconds.
- 10-bit HDR output, a two-minute video extension function and API support are listed as coming soon; API pricing is not announced.
Why it matters
Kling 4.0 stretches single-pass clips to 30 seconds and adds stereo sound and lip sync in more than ten languages, so one generation can now carry more of a finished scene. The Decoder also notes that, in its own comparison, the model holds its own against Wan 3.0 and Seedance 2.5 without pulling ahead, and suggests this could be an early sign that the rapid pace of AI video progress is levelling off. That is the outlet's speculation, and it says the next Wan and Seedance releases will tell more.
Who it affects
The update is aimed at people who make video from references and templates: the Omni Reference and Creative Recreation features target work with storyboards, existing footage, and swapped products or characters. Multilingual lip sync and in-video text matter to anyone producing content in languages such as German, English, Japanese, Korean or Spanish. Access through MCP and CLI is relevant to teams that want to build automated workflows with AI agents.
How to use it
The main model offers up to 30 seconds per pass, up to ten keyframe images to fix transitions and story beats, and up to 15 reference assets through Omni Reference. Kling 4.0 Flash trades quality limits for speed and cost, topping out at 720p and 20 seconds. The creation interface has been redesigned, and MCP and CLI access allows scripted, agent-driven workflows. API support is listed as coming soon, with pricing not yet announced. No pricing is given for Kling 4.0 or Kling 4.0 Flash.
How solid is it
Most of the feature list is the company's own description: the Omni Reference limits, the lip sync coverage, the Creative Recreation claim and the idea that Flash should close the gap with professional video are all attributed to Kling AI. The only outside assessment is The Decoder's qualitative video comparison with Wan 3.0 and Seedance 2.5; no benchmark numbers are given. The source also does not say how much cheaper or faster Flash is than the main model.
Risks and caveats
10-bit HDR output at 4K and 1080p, the video extension function of up to two minutes, and API support are not yet available. The 15-asset Omni Reference total is stated alongside per-type maximums (10 images, 5 videos, 7 elements) that add up to more than 15, and the source does not explain how the two relate. Flash is limited to 720p and 20 seconds. The Decoder says Kling 4.0 does not pull ahead of its competitors, especially in logical shot sequencing, physics and natural-sounding dialogue.
“Kling 4.0 holds its own but doesn't pull ahead, especially when it comes to logical shot sequencing, physics, and natural-sounding dialogue.”
— The Decoder