Over the course of a few minutes, Gemini continually refines its analysis and browses the Internet the way you do: search, find interesting information, and then start a new search based on what you’ve learned. This process is repeated several times and when completed, a comprehensive report with key findings is generated that you can export to a Google Doc. It’s neatly organized with links to the original sources and takes you to relevant websites and companies or organizations you might not have found otherwise, making it easy to dive deeper to learn more. If you have any additional questions about Gemini or would like to refine the report, just ask! That means hours of research available to you in just a few minutes.

Earlier this year we shared our vision to integrate more agent functionality into our products; Deep Research is the first feature in Gemini that brings this vision to life. We’ve developed a new agent system that leverages Google’s expertise in finding relevant information on the web to guide Gemini’s browsing and research. Combined with the Gemini model’s advanced reasoning capabilities and our 1M token context window, this creates comprehensive reports with helpful, easy-to-read insights.

Deep Research is the perfect tool if you are an entrepreneur starting a small business and want to quickly get competitor analysis and recommendations for suitable locations, or if you are a marketer researching current AI-powered marketing campaigns to compare for planning in 2025.

Deep Research launches today in Gemini Advanced (starting in English) on desktop and mobile web and will be available on the mobile app in early 2025. To get started, toggle the model dropdown to “Gemini 1.5 Pro with Deep Research” and enter your research question.