
Open Deep Research
β Text source-checked Nov 3, 2026An AI-powered research assistant that performs iterative, deep research on any topic by combining search engines, web scraping, and large language models.
Source-checked against the vendor's site Β· rebuilt from the vendor's own pagesProfile updated Link checked How we rate
In short
Open Deep Research is a free, open-source AI agent that performs iterative deep research using search engines and LLMs to generate detailed markdown reports.
Open Deep Research is an AI-powered research assistant that performs iterative, deep research on any topic by combining search engines, web scraping, and large language models. The repository aims to provide a simple implementation of a deep research agent that can refine its research direction over time and deep dive into a topic. The project maintains a codebase size of under 500 lines of code to ensure it is easy to understand and build upon.
The tool performs deep research by iteratively generating search queries, processing results, and diving deeper based on findings. The system uses large language models to generate targeted search queries based on research goals and previous findings. Users can configure parameters to control the breadth and depth of the research process. The assistant produces detailed markdown reports containing findings and sources.
The tool requires a Node.js environment and API keys for Firecrawl for web search and content extraction, and OpenAI for the o3 mini model. The system supports local large language models by configuring the endpoint and model name to point to a local server. The system automatically switches to use the DeepSeek R1 model instead of o3-mini when a Fireworks API key is detected.
The project is released under the MIT License, allowing users to use and modify the code as needed.
According to Open Deep Research: Open Deep Research is an AI-powered research assistant that performs iterative, Deep research on any topic by combining search engines, web scraping, and large language models. Its pages list this strength: The tool supports local LLMs via configurable endpoints and model names. They also state this limitation: Users with a free version of Firecrawl may encounter rate limit errors, requiring them to reduce the concurrency limit to 1, which slows down execution.
β The AI Tool Bible editorial team
Pros
- β The tool supports local LLMs via configurable endpoints and model names.
- β Configurable parameters allow users to control the breadth and depth of the research.
- β The system handles multiple searches and result processing in parallel for efficiency.
Cons
- β οΈ Users with a free version of Firecrawl may encounter rate limit errors, requiring them to reduce the concurrency limit to 1, which slows down execution.
Facts
- Platforms
- web, cli (not source-checked)
- Pricing
- free - Free (MIT); bring your own Firecrawl + LLM API keys (not source-checked)
- Open source
- Yes (license: MIT)
- GitHub stars
- 19,893 (checked Nov 13, 2026) (not source-checked)
- Last GitHub push
- Apr 11, 2026 (not source-checked)
- First commit
- Feb 2025 (not source-checked)
- Company
- Duet (not source-checked)
- Model used
- o3-mini (default), DeepSeek R1, or any OpenAI-compatible model (not source-checked)
Where this tool sits
Use cases
Frequently asked
- How much does Open Deep Research cost?
- The software is free under the MIT License. However, you must bring your own API keys for Firecrawl and an LLM provider like OpenAI or DeepSeek to run the tool.
- Which AI models does it support?
- It defaults to o3-mini but supports DeepSeek R1 and any OpenAI-compatible model. It can also connect to local LLMs by configuring the endpoint and model name to point to a local server.
- What are the technical requirements to run it?
- You need a Node.js environment and valid API keys for Firecrawl (for web search and extraction) and an LLM provider. The codebase is under 500 lines, making it easy to understand and build upon.
- What kind of output does it produce?
- The assistant produces detailed markdown reports containing findings and sources. It refines its research direction over time by iteratively generating search queries and processing results based on previous findings.
- Can I use it for specific research tasks?
- Yes, it is designed for deep research on any topic, iterative research direction refinement, and generating detailed markdown reports. Users can configure parameters to control the breadth and depth of the research process.