On Sunday, OpenAI unveiled a new artificial intelligence (AI) agent within ChatGPT, capable of independently conducting research on the Internet. Known as Deep Research, this tool should not be confused with Google’s similarly named AI agent powered by Gemini. According to OpenAI, the new agent is equipped to handle complex topics through multi-step research. Currently, it is accessible to subscribers of ChatGPT Pro, the company’s premium subscription level. This announcement follows closely on the heels of Google’s launch of its own Deep Research agent for the Gemini Android app.
OpenAI Unveils Deep Research AI Agent
In a blog post, the San Francisco-based AI studio outlined the capabilities of its latest research-focused tool. OpenAI claims that Deep Research can execute tasks that would typically take human researchers several hours in just a matter of minutes. This new agent complements another AI tool from OpenAI called Operator, which autonomously handles online tasks such as ticket bookings, restaurant reservations, and product purchases.
Deep Research in ChatGPT
Photo Credit: OpenAI
OpenAI emphasized the scenarios where Deep Research will be particularly advantageous, noting that it is tailored for users engaged in intensive research across fields such as finance, science, policy, and engineering. Additionally, it targets users seeking “hyper-personalized recommendations” for significant purchases typically requiring extensive research, including automobiles, furniture, and household appliances.
For ChatGPT Pro users, the Deep Research feature can be accessed via a button located beneath the text field next to the Search icon. Users can initiate the agent by submitting their query and tapping the corresponding icon. The agent will then ask follow-up questions to gather more information regarding the user’s preferences and specific needs. Users have the option to upload a file to enrich the context further.
Once users answer the follow-up queries, the AI agent will commence its research, displaying a progress bar during the process. A sidebar will provide a summary of the steps taken and the sources utilized in the research. The entire output process may take between five and thirty minutes, culminating in a comprehensive report delivered to the user once the research concludes.
This AI agent is powered by a variant of the o3 model and will include citations for the information drawn from specific sources. Moreover, the reports generated will utilize tables and sub-headings to enhance readability. OpenAI has indicated that it plans to introduce support for images, data visualizations, and other analytical outputs in these reports within the coming weeks.
At present, Deep Research is exclusively available to ChatGPT Pro subscribers, although OpenAI has announced that it will soon extend access to Plus and Teams subscription tiers. The Pro subscription is priced at $200 per month, approximately Rs. 17,500.
