OpenAI has officially released a new tool called deep research, designed to produce thorough, multi-source reports on complex topics. The feature is now available to all ChatGPT Pro subscribers, with broader access planned in the coming weeks.
How deep research works
Unlike standard chatbot responses that generate answers from a single pass of information, deep research performs a multi-step process. It browses the web, collects data from multiple sources, and synthesizes findings into a structured report. The tool uses a version of OpenAI’s reasoning model, o3, which has been optimized for extended tasks requiring analysis and synthesis.
OpenAI describes deep research as a tool for knowledge workers in fields like finance, science, policy, and engineering. It can produce reports on market trends, scientific literature, or competitive analysis. The system is designed to handle questions that would normally take a human hours or days to research.
During testing, OpenAI reported that deep research achieved a score of 26.6 percent on Humanity’s Last Exam, a benchmark designed to test AI on expert-level questions. That score is significantly higher than the 9.8 percent achieved by OpenAI’s earlier o1 model. The company also noted that deep research performed well on GAIA, a benchmark for general AI assistants, where it set a new state of the art in several categories.
Limitations and best use cases
OpenAI cautions that deep research is not perfect. It can sometimes fabricate information or make incorrect inferences, a common issue known as hallucination in large language models. The tool may also struggle with nuanced data, such as distinguishing authoritative sources from unreliable ones. OpenAI advises users to verify critical facts from the generated reports.
The tool is best suited for tasks that require breadth rather than depth. For example, it can quickly summarize the current state of research on a medical topic or provide a competitive landscape for a startup. However, it is not designed for creative writing or tasks that demand personal opinion.
Deep research is accessible through the ChatGPT interface on both desktop and mobile. Subscribers to the ChatGPT Pro tier, which costs $200 per month, can access the feature starting today. OpenAI plans to roll it out to Plus and Team users later this month, followed by Enterprise subscribers.
What this means for the industry
The launch of deep research signals a shift toward more autonomous AI agents that can perform multi-step tasks without constant human guidance. Competitors like Google have also introduced deep research capabilities in their Gemini models, indicating a broader trend in the AI industry toward tools that can handle complex, real-world research tasks.
For professionals who regularly conduct in-depth research, deep research could save significant time. Instead of manually searching through dozens of sources, a user can offload the initial gathering and synthesis to the AI. The tool provides citations and links to sources, allowing users to follow up on specific claims.
OpenAI has already published several examples of deep research in action, including reports on coffee consumption and AI chip competition. The company says the tool will improve over time as OpenAI collects feedback from early users.
For those looking to stay ahead of these developments, check out our analysis of how AI is reshaping the workplace. Read more about AI workplace trends.
