Open Source Tool arxiv_assistant: How Large Language Models Help Researchers Filter Papers Efficiently

arxiv_assistant is an open-source research assistant tool powered by Large Language Models (LLMs), designed specifically to solve the time-consuming problem of filtering ArXiv papers for researchers. By deeply understanding users’ professional backgrounds and interests, this tool intelligently recommends highly relevant latest papers. Users can jump to CoolPaper with one click for online assisted reading, supporting paper collection and categorized management, and providing offline import/export and cloud synchronization features (via WebDAV). The project is open source with an online demo available. It’s recommended to use with high-speed large models like GPT-OSS-120B. Future plans include optimizing ArXiv access proxy and custom prompt engineering. This tool aims to enhance research efficiency, and feedback from trials is welcome.