open Deep Research – free the AI agents!

OpenAI just published Deep Research, a very nice assistant that can perform deep searches on the web to answer user questions.

However, their agent has a huge downside: it’s not open. So we’ve started a 24-hour rush to replicate and open-source it. Our resulting open-Deep-Research agent took the #1 rank of any open submission on the GAIA leaderboard! ✨

OpenAI recently launched Deep Research, an advanced AI agent designed to conduct multi-step research tasks efficiently. This tool allows users to input complex queries, and it autonomously gathers, analyzes, and synthesizes information from a wide range of online sources, producing comprehensive reports akin to those generated by research analysts. The system is particularly beneficial for professionals in fields such as finance, science, and policy-making, where thorough and reliable research is critical1.

Key Features of Deep Research

• Autonomous Operation: Users provide a prompt, and Deep Research independently navigates the web to find relevant information.
• Time Efficiency: It can complete tasks that would typically take humans several hours in a matter of minutes1.
• Comprehensive Reporting: Outputs include detailed citations and summaries, making it easy for users to reference the information
• Enhanced Capabilities: It utilizes a specialized version of OpenAI’s o3 model optimized for web browsing and data analysis, enabling it to handle complex inquiries effectively1.

Open Deep Research

In addition to OpenAI’s offering, Open Deep Research has been developed by Hugging Face as an open-source alternative. This version aims to replicate the functionality of OpenAI’s system while allowing developers to contribute to its evolution. It has shown promising results on benchmarks like GAIA, achieving high accuracy in answering complex queries

Performance Metrics

• The original Deep Research achieved a 67% correct answer rate on the General AI Assistants benchmark (GAIA) under one-shot conditions and 47.6% on more challenging multi-step reasoning questions.
• Open Deep Research has also demonstrated competitive performance in similar evaluations, indicating its potential as a robust research tool3.

結論

Both Deep Research and its open-source counterpart are significant advancements in AI-driven research capabilities, enabling users to streamline their information-gathering processes while ensuring accuracy and depth in their findings.