Cool Papers: An AI-Powered Academic Paper Discovery Platform by Su Jianlin

Forum topic · 小凯 · 2026-03-06

Summary

Cool Papers (papers.cool) is a free, AI-driven academic paper discovery platform developed by Su Jianlin, the blogger behind Science Space. Built on Kimi Chat AI (Moonshot AI), it provides interactive paper explanations and currently indexes nearly 170,000 papers from arXiv and top conferences such as NeurIPS, ICML, and CVPR. The platform features full-text search powered by the Whoosh retrieval library with BM25 ranking, daily automatic updates of new papers, and a Chinese-friendly interface and AI interpretation. The tech stack includes the BottlePy web framework, Python's built-in Shelve key-value database, and the arXiv API. Launched in early 2024, the site reached 170,000 indexed papers by May 2024. Compared to tools like Semantic Scholar, ReadPaper, and Elicit, Cool Papers focuses on fast discovery plus AI interpretation rather than deep reference management, making it best suited for researchers who need a daily reading entry point for computer science and AI literature.

Cool Papers: AI-Powered Academic Paper Discovery Platform

Official website: https://papers.cool/

Cool Papers is an AI-driven academic paper discovery platform developed by Su Jianlin (author of the Science Space blog). It provides interactive AI-based paper interpretation powered by Kimi Chat AI.

Key Features

| Feature | Description | |---------|-------------| | Paper coverage | arXiv + top conference papers, nearly 170,000 papers currently | | AI interpretation | Interactive FAQ powered by Kimi Chat AI | | On-site search | Full-text search (Whoosh retrieval library + BM25) | | Daily updates | Automatic ingestion of the latest papers | | Free to use | No registration required | | Chinese-friendly | Both UI and AI interpretation support Chinese |

Technical Architecture

Backend: BottlePy (lightweight Python web framework)
Database: Shelve (Python built-in key-value database)
Search engine: Whoosh (Python full-text search library) + BM25 similarity
AI capability: Kimi Chat AI (Moonshot AI large language model)
Data sources: arXiv API + conference paper crawling

Timeline

Early 2024: Project launched and opened to the public
May 2024: Indexed papers reached 170,000 (80,000 arXiv + 80,000 conference papers); on-site search went live
2025: Continuous updates, improving search experience and AI interpretation

Use Cases

1. Daily paper browsing: View the latest arXiv submissions 2. Quick paper comprehension: Ask Kimi AI questions to rapidly grasp a paper's core content 3. Keyword search: Find related research via full-text retrieval 4. Conference tracking: Follow the latest results from top conferences (NeurIPS, ICML, CVPR, etc.)

Comparison with Similar Tools

| Tool | Paper count | AI interpretation | Search | Highlights | |------|-------------|-------------------|--------|------------| | Cool Papers | 170,000+ | ✅ Kimi | On-site full-text search | Chinese-friendly, free | | Semantic Scholar | 200M+ | ❌ | Semantic search | Citation network, impact analysis | | ReadPaper | Unknown | ✅ | Basic search | Translation, knowledge graph, reference management | | Elicit | Unknown | ✅ | Research Q&A | Data extraction, research question exploration |

Limitations

Discipline coverage: Focused mainly on arXiv and computer science, not all disciplines
Search capability: Based on BM25; semantic understanding is limited compared to vector retrieval
Feature depth: Emphasizes quick browsing; weaker on in-depth reference management

Who It's For

Researchers who need to quickly scan daily arXiv updates
Readers who want AI assistance to understand the core content of papers
Developers following the latest advances in computer science and AI

Key Insight

Cool Papers has a clear positioning: "fast discovery + AI interpretation", not comprehensive reference management. It works well as a daily paper browsing entry point, paired with other tools (such as Zotero or ReadPaper) for deep reading and reference management.