Deep Research in ChatGPT Gets New Features
In its release notes for ChatGPT, the AI giant announced and detailed the new capabilities in Deep Research. The improvements can be divided into two groups: the first focusing on the new controls, and the second on the viewing experience. In the first group, OpenAI has now added an option to let users narrow the focus of research to specific websites.
This was a highly requested feature because, so far, the only way Deep Research could be grounded was by specifying the website limiters in the prompt itself. However, the ability to add this information into the tool at a system level will let users see more accurate and focused responses. Additionally, users can also enable connected apps while conducting research and use them as trusted sources.
Apart from this, users can also edit a research plan before it begins. This is particularly useful given that earlier users could only redirect the research after it started. If a user wanted to alter details before the plan was created, they would have to start all over again. The ability to adjust the direction of the research mid-run continues to be available.
Coming to the second group of improvements, Deep Research now has a redesigned sidebar entry point and a full-screen report viewer. The latter opens in a document viewer-like interface, complete with side panels for index (on the left) and citations (on the right). The index is also interactive, and users can click on any heading or subheading to quickly reach that part.
Separately, the AI firm has also updated the GPT-5.2 Instant model in ChatGPT and application programming interface (API). OpenAI says the responses are now “more measured and grounded” in tone, which makes them contextually appropriate to the conversation. Further, the updated model also provides clearer output and more relevant answers to how-to questions and advice.
