STORM

STORM is an open-source knowledge curation system from Stanford’s OVAL lab that uses large language models to research a topic on the web and generate a full-length, citation-backed report that reads like a well-structured Wikipedia article. Give it a subject, and it plans the research, gathers sources, and writes the article for you.

Overview

STORM (Synthesis of Topic Outlines through Retrieval and Multi-perspective Question Asking) has earned more than 31,600 stars on GitHub and was presented at NAACL and EMNLP 2024. Its core insight is that good articles come from good questions: STORM simulates conversations between a writer persona and a topic expert grounded in web sources, using multi-perspective question asking to surface angles a single query would miss. It then organizes the findings into an outline and writes a complete report with references. A collaborative mode, Co-STORM, lets humans join the process, steer the research direction, and refine the shared knowledge base as it grows.

Key Features

• End-to-end pipeline that turns a single topic into a structured, full-length report
• Multi-perspective question asking that explores a topic from several viewpoints
• Automatic citations linking claims back to the web sources they came from
• Co-STORM mode for human-AI collaborative research and knowledge curation
• Works with multiple LLM providers and search or retrieval backends
• Modular Python design (DSPy-based) that researchers can extend and customize
• Free web demo hosted by Stanford for trying the system in a browser

💳 Pricing

STORM is free and open source under the MIT license. Running it yourself requires your own LLM API key and, depending on configuration, a search API key.

Who Is It For

STORM is ideal for researchers, writers, students and analysts who need a fast, well-sourced first draft on an unfamiliar topic, and for developers building deep-research or report-generation features into their own products.