With the emergence of Sora, text-to-video models have become a new focus of competition among Chinese tech companies. Kuaishou's "Kelin" AI model, which uses a similar technical approach to Sora, can generate high-definition videos up to 2 minutes long, and currently has over a million people applying for internal testing. Other companies such as Zhipu AI, Tencent, and ByteDance are also following suit, launching their own text-to-video models.
Kelin AI has undergone three iterations in less than two months since its launch, with significant improvements in generating details, composition, and camera movements. It can produce videos with 1080p resolution and up to 2 minutes in length, surpassing most similar products on the market. Users generally rate Kelin AI positively, considering its results "far exceeding expectations."
Kuaishou CEO Cheng Yixiao initiated a new AI strategy last year. From project initiation in March to launch in June, Kelin AI was developed in just 3 months. This was made possible by Kuaishou's resource allocation and technological innovation, including adopting a Sora-like technical approach combined with proprietary technology.
Major tech companies are entering the text-to-video field, partly due to huge market demand and partly to create commercial value by applying large models. Kuaishou's lead is mainly attributed to its massive video data accumulation, hardware cooperation with NVIDIA, and full company support.
However, the technological gap in the text-to-video field is not significant, and future competition will mainly focus on computing power scale and training data. Kuaishou is not the strongest in computing power, so whether it can maintain its lead remains to be seen. Currently, domestic companies are generally adopting a strategy of launching products first and then optimizing them to capture market share.
The rapid growth of Kelin AI users is partly due to user curiosity and partly due to traffic support from the Kuaishou platform. Kelin AI has already launched a paid membership system, beginning to explore commercialization. In the future, Kuaishou hopes to build a content ecosystem through Kelin AI, empowering creators and improving overall platform content quality.
Overall, text-to-video models are becoming a new competitive hotspot in the AI field, with various companies entering the market. Kuaishou is temporarily leading with Kelin AI, but future competition will remain intense. This technology is expected to bring revolutionary changes to content creation and is worth continued attention.