How Deep Research works: now you can see how the AI reasons

OpenAI has launched Deep Research, an agent integrated into ChatGPT with extensive reasoning capabilities that allow it to delve into the web and analyze and synthesize large amounts of data to conduct thorough investigations.

Deep Research is now available in the Pro plan, allowing for up to 100 queries per month. The developer has announced that this capability will soon be available for the Plus and Team plans. Moreover, they state that they are working to provide access to users in the United Kingdom, Switzerland, and the European Economic Area.

Today we are launching our next agent capable of doing work for you independently—deep research.

Give it a prompt and ChatGPT will find, analyze & synthesize hundreds of online sources to create a comprehensive report in tens of minutes vs what would take a human many hours. pic.twitter.com/03PPi4cdqi

— OpenAI (@OpenAI) February 3, 2025

An agent driven by the reasoning capabilities of o3

Deep Research has been designed to work independently on complex research tasks. «You give it a prompt and ChatGPT will find, analyze, and synthesize hundreds of online sources to create a comprehensive report at the level of a research analyst», they explain from OpenAI.

Thus, Deep Research will complete the task in several steps and will take between 5 and 30 minutes, whereas it would take a person many hours for such deep research. This AI agent allows savings in time to invest it in other tasks, freeing users from the tedious collection of data and sources.

Deep Research is powered by a version of o3, OpenAI’s most advanced reasoning model, and is trained on real-world tasks requiring the use of a browser and a Python tool, using reinforcement learning.

«Through that training, it learned to plan and execute a multi-step trajectory to find the data it needs, backtracking and reacting to information in real-time when necessary.

The model can also explore files uploaded by the user, chart and iterate graphs using the Python tool, embed generated graphs and images from websites in its responses and cite specific sentences or passages from its sources. As a result of this training, it reaches new heights in a series of public evaluations focused on real-world problems».

How Deep Research works

This AI agent can be very useful both in the professional or work environment and in personal life. On one hand, you can rely on it for complex projects in areas such as finance, politics, engineering, or science that require large-scale, accurate and reliable research. On the other hand, you can use it to conduct a careful market study when planning an important purchase, such as a car, an appliance, or an electronic device, for instance.

To use Deep Research, you only need to access ChatGPT and select the “Deep Research” option located below the chatbot’s text box. After this, provide precise instructions to explain to the AI the task it must perform and, if necessary, attach files or spreadsheets.

Before starting its information search process, the AI might ask you for details about the task to refine its results further. Once it begins conducting research, you will see a loading bar below the chatbot’s last message, and a new view will open on the right side where Deep Research will show you the steps it is executing and the sources it is using.

Once it has completed its research, you will receive a notification and the loading bar will appear complete. Deep Research will create a final report shown within the chat, which will include links to the pages and websites from which it has taken the data. OpenAI has reported that, in the following weeks, it will add embedded images, data visualizations, and other analytical possibilities to offer clearer reports with more context.

The race for General AI

OpenAI’s grand objective, shared by other major AI companies, is to achieve the creation of  Artificial General Intelligence (AGI). This technology, which for now is only a concept floating in the realm of ideas, would be capable of performing tasks that require human intelligence and could even surpass our cognitive versatility.

An ambitious goal that, although still far from becoming a reality, progresses with each new product launched by major technology developers of AI.

«The ability to synthesize knowledge is a prerequisite for creating new knowledge. For this reason, Deep Research represents an important step towards our broader objective of developing general artificial intelligence, which we have long envisioned as capable of producing new scientific research», states OpenAI.

Photo: OpenAI