Artificial intelligence, represented by it, is permeating every corner of people's work and life, and the scientific research field is no exception. Recently, Alibaba released a hundred billion parameter level quark model independently developed based on the Transformer architecture. It is reported that this large model can be used for scientific research data collection, rapid literature reading and translation, creative polishing and other scenarios. Not only Alibaba, but also companies such as iFlytek and Tencent have launched large model products to assist scientific research. The emergence of this series of products is quietly changing the way scientific researchers work. The big model has entered the field of scientific research earlier this year, and the popularity of ChatGPT has sparked a wave of language big models. People can ask questions as they please, and big models always provide answers. This trend quickly spread to the field of scientific research. Shortly after the release of ChatGPT, an application plugin called txyz.ai gained popularity in the research community. This is a plugin designed specifically for reading scientific literature, leveraging the powerful comprehension capabilities of ChatGPT. Users can directly upload the full text of their paper to the application and provide corresponding interpretation requirements, so that it can answer the questions raised by users at the fastest speed. Even if the full text of the paper is not downloaded, it doesn't matter. txyz.ai supports searching the preprint website of the paper. Users can provide only one paper ID number, and txyz.ai will automatically search and learn the paper, and provide answers according to user needs. Not only that, users can also have conversations with them through chatting and ask various questions about the content of the paper. Yan Bojun, a well-known science popularization blogger on the video website Bilibili, is a loyal user of txyz.ai. He often uses txyz.ai in various science popularization videos to assist in interpreting papers. In his view, the accuracy of the paper interpretation provided by txyz.ai is already very high and has strong practical application value, which can help researchers search and read literature more efficiently. "There are certainly situations where txyz.ai cannot be interpreted," Yan Bojun said. Based on his usage experience, the questions posed to txyz.ai must be a "valid question" that can be answered. "If the question is asked too finely, too deeply, or too tricky, it will directly tell you that it cannot be answered.". However, it must be acknowledged that in today's rapidly developing and gradually entering various industries, there are still not many large model products specifically designed for the scientific research field, and most of them are experimental products. Not long ago, when iFlytek released the latest version of iFlytek's Spark Cognitive Model V3.0, it also released 12 industry-specific large models in one go. Among them are the scientific and technological literature model for scientific researchers jointly developed by the Literature and Information Center of the Chinese Academy of Sciences, and the application product based on this model - Spark Research Assistant. This is also one of the few large-scale model products specifically launched for scientific research in China. At present, the Spark Research Assistant has three major functions: achievement research, paper reading, and academic writing. The library staff of the First Affiliated Hospital of Zhejiang University spoke in English