Xiaodigua: Xiaohongshu's Self-Developed LLM 115

Xiaohongshu's self-developed large model "Xiaodigua" was developed by the large model team led by Zhang Debing (RedNote name: Yuchen), the head of Xiaohongshu's AI innovation. This self-developed general-purpose large model base has already undergone gray-scale testing on some internal products. The exploration of Xiaohongshu's AI products is mainly led by Deng Chao (RedNote name: Yingmu), the head of Xiaohongshu's product and design.

Xiaohongshu plans to apply AI applications in directions and scenarios such as multimodal technology, AI content creation tools, social, and search. Xiaohongshu is very cautious about launching new AI products, because it is worried about damaging the existing content ecosystem. Xiaohongshu's content ecosystem is built on real-life recommendations, and the role and positioning of AI within it is something they need to consider. Social and search, on the other hand, represent new scenarios where Xiaohongshu hopes to step out of its comfort zone of image-and-text recommendations and reach more users with AI.

Xiaohongshu hopes to use AI technology to break through the limitations of its existing image-and-text content and further expand into new scenarios such as social and search, in order to attract more users.

In September 2023, Xiaohongshu first launched the generative search dialogue assistant "DAVINCI" based on the large language model from the large model company MiniMax, and conducted internal testing. On October 24 of the same year, Xiaohongshu had already applied to register the trademark "Xiaodigua" for its self-developed large model. The international classification of this trademark is scientific instruments, and it is currently in the registration application stage. Multiple insiders indicated that several AI products have already been connected to Xiaodigua's API for testing. Xiaodigua's performance in applications has reached the level of mainstream domestic large models.