ChatGPT has introduced a feature called Deep Research that performs multi-step online research. The tool uses a version of the upcoming OpenAI o3 model made for web browsing and data analysis. It looks for information from various online sources including text, images and PDFs and then creates a detailed report with clear citations. This report explains the process of reasoning used to reach the conclusions. The feature is designed for professionals in areas such as finance, science, policy and engineering as well as for consumers who need verified advice on products like cars and home appliances.
The Deep Research tool works much like a human research analyst. It uses reinforcement learning methods to plan its search steps and can go back to gather more information if needed. The tool is also able to use Python to create graphs and visuals, which are included in its reports. Tests using benchmarks like Humanity’s Last Exam and GAIA show that this tool performs with higher accuracy compared to earlier models. Although it may sometimes have minor formatting issues or make small errors, its output remains clear and reliable. At present, the feature is available for Pro users with a limit of 100 queries per month.
ChatGPT Deep Research is now available on the web and there are plans to launch it on mobile and desktop platforms soon. OpenAI is working on a faster and more cost-effective version that will support higher usage in the future. The feature is compute intensive and research tasks may take between five and thirty minutes to complete. OpenAI continues to check its performance and will improve the service through more testing and updates, ensuring that users receive clear, well-documented research reports.
