Can be built into companies' own products
China's Kuaishou said on Kling's official X account on Oct. 21 (early Oct. 22 Japan time) that it would begin offering an API for its video generation AI "Kling 4.0." It is accepting applications for use and will open access to businesses and developers around the world in stages. With the API, companies can build Kling's video generation into their own products and workflows.
Kling 4.0 was released on Oct. 15. Until now, all of the model's features were available only in the company's own web app, and it had also provided a tool for calling it from AI agents.
Up to 30 seconds per generation, up to 15 reference materials
Kling 4.0 can create videos of up to 30 seconds per generation. In addition to creating videos from text or images, users can specify the first and last scenes or line up as many as 10 keyframe images. Up to 15 materials in total can be used as references, with up to 10 images and up to 5 videos, and voices can also be specified. Audio comes in two channels, left and right.
The official documentation also lists 4K resolution, but it is currently marked "coming soon," and the resolutions actually available are 720p and 1080p.
fal and other outside services add support at once, pricing undisclosed
U.S.-based fal, which provides AI models, opened Kling 4.0 to all its users on Oct. 23. More than 10 services, including China's Meitu and U.S.-based Higgsfield and OpenArt, integrated it at the same time.
However, as of Oct. 23, the pricing of the Kling 4.0 API, the model identifier and usage limits had not been disclosed. The model is still labeled "preview," and no timing has been given for an official release. Available features may differ depending on the outside service.
