Mario Nawfal

1. DeepResearch is OpenAI’s latest AI agent that works independently to handle research tasks. Just give it a prompt, and ChatGPT will search, analyze, and summarize information from hundreds of online sources delivering a detailed report like a research analyst. https://x.com/OpenAI/status/18...

2. DeepResearch is designed for professionals in fields like finance, science, policy, and engineering who need thorough, accurate, and reliable research. It’s also great for shoppers looking for personalized recommendations on things like cars, appliances, or furniture items that usually require careful research. https://x.com/OpenAI/status/18...

3. How to Use DeepResearch In ChatGPT, select "DeepResearch" in the message composer and enter your query. Describe what you need—whether it’s a competitive analysis on streaming platforms or a personalized report on the best commuter bike. You can also attach files or spreadsheets to provide context. Once it starts running, a sidebar will appear, summarizing the steps taken and the sources used. DeepResearch typically takes 5 to 30 minutes, depending on the complexity of the request, as it thoroughly searches the web. You can step away or work on other tasks—you’ll receive a notification when it’s done. The final output arrives as a detailed report within the chat. Over the next few weeks, OpenAI will also add embedded images, data visualizations, and other analytic outputs to make reports even clearer and more informative.

4. Powered by a specialized version of OpenAI’s o3 model, optimized for web browsing and Python-based analysis, DeepResearch uses advanced reasoning to scan and interpret text, images, and PDFs from across the internet.

5. The model behind DeepResearch achieves new benchmarks on several public evaluations designed for real-world problem-solving, including Humanity’s Last Exam.

6. On GAIA, a public benchmark for real-world AI evaluations, the model behind DeepResearch achieves state-of-the-art (SOTA) performance, ranking at the top of the external leaderboard. GAIA tests AI across three difficulty levels, evaluating key skills like reasoning, multi-modal fluency, web browsing, and tool usage.

7. Limitations Deep research unlocks significant new capabilities, but it’s still early and has limitations. It can sometimes hallucinate facts in responses or make incorrect inferences, though at a notably lower rate than existing ChatGPT models, according to internal evaluations. It may struggle with distinguishing authoritative information from rumors, and currently shows weakness in confidence calibration, often failing to convey uncertainty accurately. At launch, there may be minor formatting errors in reports and citations, and tasks may take longer to kick off. We expect all these issues to quickly improve with more usage and time.

8. DeepResearch is available today on ChatGPT web and will be rolled out to mobile and desktop apps within the month. Currently, it can access the open web and any uploaded files. In the future, you’ll be able to connect to more specialized data sources, including subscription-based or internal resources, making DeepResearch’s outputs even more robust and personalized.