New OpenAI 'Deep Research' Agent Turns ChatGPT into a Research Analyst
- By John K. Waters
- 02/12/25
OpenAI has introduced a new "Deep Research" feature that enhances ChatGPT with the capabilities of a "research analyst" that automates time-consuming research by retrieving, analyzing, and synthesizing online information.
Unlike standard chatbot interactions, Deep Research operates independently for 5 to 30 minutes, browsing the web, interpreting content, and compiling structured reports with citations, the company said. Powered by a specialized version of OpenAI's upcoming o3 model, it's optimized for reasoning and data analysis. OpenAI is targeting professionals in finance, policy, and engineering with this release, as well as users seeking comprehensive insights on complex topics.
OpenAI emphasized the tool's accuracy, citing an unprecedented 26.6% score on "Humanity's Last Exam," a benchmark designed to test expert-level reasoning across 100 subjects. In contrast, its predecessor, GPT-4o, scored 3.3%, and Google's Grok-2 achieved 3.8%.
However, the company acknowledged ongoing challenges, including occasional inaccuracies and difficulties distinguishing authoritative information from rumors. Verification by users remains critical, according to experts, given AI's tendency to "hallucinate" or fabricate information.
This launch follows closely on the heels of advances by competitors like Google's Deep Research, introduced to Gemini Advanced subscribers in December 2024, and Chinese startup DeepSeek's claims of superior efficiency. OpenAI described its tool as a "significant step" toward its vision of artificial general intelligence.
Although the Deep Research agent is optimized for professionals in finance, science, and engineering, the company says it can also assist with consumer-focused decisions, such as comparing cars or furniture. It accepts diverse inputs, including spreadsheets, images, and PDFs, to provide tailored responses, complete with bullet points, tables, and charts in future updates.
Deep Research is currently available to Pro subscribers at $200 per month, with a limit of 100 queries. OpenAI plans to extend access to Plus, Team, and Enterprise tiers, alongside enhanced rate limits as faster iterations are developed. The feature is web-only but will expand to mobile and desktop platforms within the month.
The launch underscores increasing competition in the AI sector, with rivals such as Google and DeepSeek innovating rapidly. Google's multi-step research agent has drawn comparisons to OpenAI's offering, though it remains in limited rollout.
In an interview with The Guardian, Andrew Rogoyski, from the Institute for People-Centred AI at the University of Surrey, cautioned against relying solely on AI outputs without rigorous checks: "There's a fundamental problem with knowledge-intensive AIs and that is it'll take a human many hours and a lot of work to check whether the machine's analysis is good."
About the Author
John K. Waters is the editor in chief of a number of Converge360.com sites, with a focus on high-end development, AI and future tech. He's been writing about cutting-edge technologies and culture of Silicon Valley for more than two decades, and he's written more than a dozen books. He also co-scripted the documentary film Silicon Valley: A 100 Year Renaissance, which aired on PBS. He can be reached at [email protected].