OpenAI Launches ChatGPT Deep Research

Published:February 3, 2025

Reading Time: 3 minutes

OpenAI has introduced a new feature for ChatGPT called ‘Deep Research’. This tool is targeted at professionals who rely on thorough, complex research in their day-to-day work. Disciplines in STEM, Finance, and Political science stand a lot to gain from this new feature. Deep Research works by condensing traditional research time and turning up a handful of well-sifted resources with verifiable sources.  

It’s also ideal for anyone who needs to make important purchases that require careful evaluation, like cars or appliances.

What Is ChatGPT Deep Research?

ChatGPT Deep Research isn’t just another tool for quick answers. It was made to deep into your questions, sifting through multiple sources to provide you with a comprehensive, reliable response. This is perfect for situations where you need to pull together information from various perspectives and sources, not just a quick summary.

The tool’s goal is to replace traditional research, condensing the time you would normally spend looking for credible sources. But without sacrificing thorough and actionable results.

How Does ChatGPT Deep Research Work?

Using ChatGPT Deep Research is easy. Simply select the “deep research” option in the ChatGPT composer, type in your query, and attach any relevant files or spreadsheets if needed. This feature is currently available through the web but we can expect mobile and desktop app integrations soon. 

Once you’ve submitted your query, the research process takes anywhere from 5 to 30 minutes, depending on the complexity of the task. When the research is complete, you’ll receive a detailed, text-only answer (for now) that’s fully documented with citations and a clear summary of how the information was gathered. 

This transparency is crucial, as it allows you to verify the data and check the sources. ChatGPT Deep Research is available to ChatGPT Pro users first, with plans to expand to Plus, Team, and eventually Enterprise users. The rollout is targeted by region, with some countries still awaiting access.

A Reliable and Documented Research Tool

What makes ChatGPT Deep Research stand out is its documentation. Every response will be backed by citations, and OpenAI will include a breakdown of the reasoning behind the conclusions. This means you can cross-check the information and verify the results with ease. It’s a far cry from the quick summaries you might get from other tools that don’t show their sources.

While the tool is still new and AI is prone to occasional errors, OpenAI is committed to improving its accuracy. The team is leveraging its special o3 “reasoning” AI model, which has been trained to process large amounts of data from multiple sources and generate results that are more precise and reliable.

The Technology Behind ChatGPT Deep Research

At the heart of this feature is OpenAI’s o3 AI model. This advanced model is built specifically for web browsing and data analysis, and it uses reasoning to interpret complex information from texts, images, and PDFs. The model is trained with reinforcement learning, which helps it improve over time by rewarding it for making better decisions.

This makes ChatGPT Deep Research more efficient at pulling together reliable information from various sources and ensuring that the results are tailored to the needs of the user. As the model continues to learn and adapt, the quality of its research should improve as well.

Accuracy and Testing

OpenAI has tested ChatGPT Deep Research using Humanity’s Last Exam, a rigorous evaluation that includes over 3,000 expert-level questions from different fields. While the o3 model achieved an accuracy rate of 26.6%, which may seem low, it’s important to note that this exam was designed to be exceptionally tough. In comparison, other AI models scored much lower, showing that the Deep Research tool outperforms its competitors.

Also read: Humanity’s Last Exam: The One Test AI Couldn’t Beat

However, the tool isn’t without its limitations. It can still make mistakes, especially when dealing with ambiguous or controversial topics, and sometimes it struggles to distinguish credible sources from unreliable ones. Additionally, formatting errors in citations and reports are possible.

The Future of AI Research Tools

One of the most exciting aspects of ChatGPT Deep Research is its potential for future improvements. OpenAI plans to add more features, such as visual elements like graphs and images to supplement the text-based responses. 

Additionally, they are working on connecting the tool to specialized data sources, including paid and internal resources, to provide even more in-depth and customized research. As these enhancements roll out, ChatGPT Deep Research could become a go-to tool for anyone needing detailed, high-quality research in a fraction of the time it would take using traditional methods.

Lolade

Contributor & AI Expert