OpenAI unveiled its latest innovation, Deep Research, on February 2, 2025, a powerful AI tool designed to perform complex research tasks by gathering and synthesizing data from across the internet. This AI tool promises to function at the level of a research analyst, providing users with comprehensive, well-synthesized reports using information from various online sources, including text, images, and PDFs. Deep Research is built on a version of OpenAI’s upcoming o3 model, optimized for web browsing and data analysis tasks, making it a significant leap forward in AI-driven research capabilities.
 

Deep Research is capable of performing tasks that require extensive context and real-time information gathering, such as uncovering hard-to-find data or answering questions that would typically take from 30 minutes to 30 days for a person to research. The tool works by using reinforcement learning techniques to plan, execute, and adjust its trajectory based on the information it collects from different sources. The output includes synthesized reports, with clear citations and an organized summary of the findings. Users can upload specific files and spreadsheets to provide additional context, making the research even more tailored to their needs.
 

However, OpenAI has acknowledged that Deep Research still has its limitations, particularly when it comes to distinguishing authoritative information from rumors and handling uncertainty. Despite this, the AI tool has achieved impressive benchmark results, including top scores in tests like Humanity’s Last Exam and GAIA. Currently, Deep Research is accessible to Pro users of ChatGPT, who can use it to generate detailed research reports directly within the platform. It is a compute-intensive tool, requiring substantial processing power, and is not yet available in the UK and the EU regions.