news

Tencent Technology Author Guo Xiaojing

On July 24, Keling AI, the large model for Kuaishou video generation, announced a new round of upgrades to its basic model and announced that it would fully open its internal testing services.

Since the Kuaishou Keling AI opened the internal beta application on June 6, the number of applicants has exceeded one million, and currently more than 300,000 people have obtained the experience qualification, which means that there are still 700,000 people waiting in line. Today, Kuaishou announced that the internal beta service will be fully open from now on. Users can get 66 "inspiration points" for free by logging in every day, which can be used to redeem the use rights of designated functions or value-added services on the platform. At the same time, Keling AI officially launched the membership system, with members divided into three categories: gold, platinum, and diamond.

Along with the launch of the membership system, Kuaishou also announced that the basic model of KeLing AI will be further upgraded, mainly in terms of picture quality and motion performance.In addition, Keling AI has previously opened up functions such as image-generated videos and video continuation, and released a web version and several new functions at the World Artificial Intelligence Conference, extending the length of image-generated videos to 10 seconds.

Kuaishou's "KeLing" model is a video generation model developed by Kuaishou's AI team. It was first released on June 6, 2024 and can generate videos up to two minutes long with a resolution of 1080p. VAEand Diffusion TransformerThe architecture performs well in simulating the characteristics of the physical world and generating video content with large and reasonable movements, which has attracted attention in the industry. Subsequently, on June 21, Keling launched the image-generated video function and the video continuation function, further expanding the boundaries of video generation, with the longest video length reaching 3 minutes. The team members come from well-known universities such as Tsinghua University, University of Science and Technology of China, and Hong Kong University of Science and Technology, and have strong technical backgrounds.