OpenAI has rolled out a major update to ChatGPT's deep research feature, introducing a fullscreen document viewer that lets users browse AI-generated reports independently from the chat window. The new interface includes a table of contents in a left-hand column for easy navigation and a right-hand column that can expand to display citations. ChatGPT Plus and Pro subscribers can now export completed reports in multiple formats including Markdown, Word, and PDF.

Released last year, the deep research mode takes extra time to break down questions into parts, gather and compare information from the web, cross-check claims, and synthesize findings into coherent answers. The latest update adds options to focus research on specific websites and trusted sources, along with a redesigned sidebar entry point. Users can now create and edit research plans before starting, track progress in real time, and adjust the research direction mid-run.

The update also includes improved accuracy controls and credibility features, allowing users to add sources while reports are being generated. OpenAI says the new viewer tool is currently rolling out to Plus and Pro subscribers, with ChatGPT Go tier and free users set to receive the update in the coming days. This enhancement aims to make the deep research experience more intuitive and manageable while maintaining the feature's core strength of producing well-supported, comprehensive reports.