Open Deep Research

An open-source, LLM-powered agent workflow that autonomously conducts multi-step research, synthesizing complex web data into comprehensive, well-cited reports.

Open Deep Research is a robust, configurable LLM agent workflow built on LangGraph, designed to execute high-level research tasks. It operates via a precise three-step process: Scope, Research, and Write, handling complex, multi-hop reasoning that traditional search fails to address. The system is benchmarked against the Deep Research Bench (100 PhD-level tasks) and is engineered for flexibility, supporting a wide range of LLM providers and search tools like the Tavily API. This architecture allows developers and users to generate in-depth, structured reports, moving beyond simple Q&A to deliver comprehensive, cited analysis on par with a human analyst.

https://opendeepresearch.dev