Recently, Zhihu announced a significant update to its AI search product, Zhihu Direct Answer, which has officially integrated the "fully powered" DeepSeek-R1 model. This initiative aims to comprehensively enhance the reasoning capabilities of Zhihu Direct Answer and provide users with a higher-quality AI search experience.
It is reported that the DeepSeek-R1 model boasts excellent logical reasoning and structured output capabilities, performing exceptionally well in text generation scenarios. Zhihu Direct Answer, on the other hand, has access to a vast amount of high-quality content within the community and over 50 million Chinese and English literature data, giving it strong professional content search and tracing capabilities. The combination of the two will bring a significant innovation to AI search products.
By integrating the DeepSeek-R1 model, the reasoning process of Zhihu Direct Answer has become more accurate and precise, based on the community's professional and traceable content. The community's professional content will be structured and outputted through this enhanced reasoning capability, leading to a comprehensive improvement in the quality of search results and answers. Users can choose to utilize the DeepSeek-R1 model for both general and professional searches when using Zhihu Direct Answer.
Additionally, Zhihu Direct Answer has simultaneously launched a knowledge base feature. This feature integrates modules for local file uploads, web bookmarking, RSS subscriptions, and Zhihu favorites, making it easier for users to build and manage their personal knowledge base. Users can employ the DeepSeek-R1 model for information extraction, analysis, and sharing of unique content such as papers, documents, web pages, and Zhihu Q&A, further enhancing the efficiency of knowledge work.