OpenAI has released a powerful agent feature that enables ChatGPT to perform complex, multi-step research tasks online. The feature, called “deep research,” reportedly takes only tens of minutes to complete tasks that would take human researchers hours or even days to complete.
OpenAI describes “deep research” as a major milestone in its journey toward artificial general intelligence (AGI).
“The ability to synthesize knowledge is a prerequisite for creating new knowledge. As such, deep research marks an important step toward our broader goal of developing AGI,” OpenAI said.
Agentic AI helps ChatGPT assist with complex research
Deep research enables ChatGPT to autonomously find, analyze, and synthesize information from hundreds of online sources. With just a prompt from the user, OpenAI says the tool can provide a comprehensive report that rivals the output of a research analyst.
With the power of OpenAI’s upcoming “o3” model variant, the model is designed to free users from time-consuming, labor-intensive information gathering. Whether it’s a competitive analysis of streaming platforms, a sensible policy review, or a personalized recommendation for a new commuter bike, deep research is guaranteed to deliver precise and reliable results.
Importantly, each output includes full citations and transparent documentation — enabling users to easily verify findings.
The tool appears particularly good at uncovering niche or non-intuitive insights, making it a valuable asset for industries such as finance, science, policymaking, and engineering. But OpenAI also envisions deep dives being useful to regular users, such as those looking for hyper-personalized recommendations or shoppers for a specific product.
This latest agent feature runs through ChatGPT’s user interface; users simply select the “drill down” option in the message editor and enter a query. Support files or spreadsheets can also be uploaded for more context.
Once initiated, the AI begins a rigorous, multi-step process that can take 5-30 minutes to complete. A sidebar provides updates on actions taken and sources consulted. Users can continue with other tasks and receive notifications when the final report is ready.
The results are presented in the chat as detailed, well-documented reports. In the coming weeks, OpenAI plans to further enhance these outputs by embedding images, data visualizations, and charts to provide greater clarity and context.
Unlike GPT-4o, which excels at real-time multimodal conversations, Deep Search prioritizes depth and detail. Its ability to rigorously cite sources and provide comprehensive analysis sets it apart—shifting the focus from quick, summary answers to well-documented, research-grade insights.
Built for real-world challenges
Deep Rsearch leverages a sophisticated training approach, grounded in real-world browsing and reasoning tasks across diverse domains. Its model is trained via reinforcement learning to autonomously plan and execute multi-step research pipelines, including backtracking and adaptively improving its approach as new information emerges.
The tool can browse user-uploaded files, generate and iterate on graphs using Python, embed generated media such as images and web pages into responses, and cite exact sentences or paragraphs from its sources. This extensive training results in a highly capable agent that can solve complex real-world problems.
