In a testament to the rapid advancements in the AI market, a team at Hugging Face has successfully replicated OpenAI’s Deep Research agent, which was released earlier this week. The new open-source initiative, named Open DeepResearch, has already matched the performance benchmarks of OpenAI’s product at a significantly lower cost.
What is Deep Research?
Deep Research is an AI agent designed to conduct complex, multi-step web research using reasoning and a base Large Language Model (LLM), specifically the unreleased o3 model. OpenAI claims that this agent is capable of adapting its research direction mid-way if new data emerges, showcasing its advanced reasoning capabilities.
According to OpenAI, “Powered by a version of the upcoming OpenAI o3 model optimized for web browsing and data analysis, it leverages reasoning to search, interpret, and analyze massive amounts of text, images, and PDFs on the internet, pivoting as needed…”
Introducing Open DeepResearch
Hugging Face’s Open DeepResearch was launched just 24 hours after the commercial release of Deep Research. The goal is to provide similar functionality for free, making it accessible to anyone worldwide.
Challenges and Goals
Replicating Deep Research involves several major challenges. The open-source agent must integrate multimodal support with reasoning and task chaining to effectively solve problems in real-time.
One of the key targets for the Hugging Face team is to achieve parity with Deep Research on the GAIA leaderboard, which validates autonomous agent frameworks. OpenAI’s Deep Research scored 67.36% on the validation benchmark, compared to just 9.7% from the bare GPT-4 Turbo LLM without agentic support.
By overcoming these challenges, Open DeepResearch aims to democratize access to advanced AI research tools, making them available to a broader audience.
