The announcement, made on Sunday, February 2, comes at a time when the AI landscape is heating up with competition, notably from the Chinese AI model DeepSeek, which has recently made waves in Silicon Valley.

Deep research is an agentic capability within ChatGPT designed to undertake multi-step research on the internet for complex tasks. With this tool, users can pose a query, and ChatGPT will independently find, analyze, and synthesize hundreds of online sources to produce a comprehensive report that matches the quality of work by a professional research analyst.

This new feature promises to accomplish in "tens of minutes what would take a human many hours."

The tool is powered by an advanced version of the upcoming OpenAI o3 model, optimized for web browsing and data analysis. It uses reasoning to navigate, interpret, and analyze vast amounts of text, images, and PDFs online, adjusting its approach based on the information it encounters.

This feature targets professionals in fields like finance, science, policy, and engineering, who require thorough and reliable research. It’s also beneficial for consumers looking for detailed, personalized recommendations for significant purchases. Each output from deep research includes clear citations and a summary of its methodology, ensuring transparency and verifiability.

To use deep research, users simply select the option in the ChatGPT interface, input their query, and can even attach files to provide additional context. The tool could take from 5 to 30 minutes to deliver its findings, during which users can continue with other tasks, receiving a notification when the research is complete. The result is presented as a detailed report within the chat, with plans to soon include embedded images and data visualizations for enhanced clarity.

The model behind deep research has set new benchmarks in public evaluations like Humanity’s Last Exam, achieving a score of 26.6% accuracy across a broad spectrum of subjects. This performance indicates significant advancements in the AI’s ability to handle real-world, domain-specific inquiries with depth and detail beyond superficial summaries.

The introduction of deep research coincides with heightened competition from DeepSeek, which has caused a stir in the tech world due to its performance and cost-effectiveness.

DeepSeek, released on January 20th, has quickly climbed to the top of Apple Store’s download charts, capturing the attention of AI experts and the entire tech industry. The model’s ascent has been nothing short of a phenomenon, astonishing investors and leading to a notable downturn in some tech stocks.

U.S. President Donald Trump has labelled DeepSeek’s rise as a "wake-up call" for American companies, urging them to focus on "competing to win" in the global AI landscape.

DeepSeek’s unique selling point is its development at a fraction of the cost compared to leading models like those from OpenAI, achieved by utilizing fewer advanced chips.

This efficiency has directly challenged the established market dynamics, causing chip-making giant Nvidia to lose nearly $600 billion in market value in just one day - marking the largest single-day drop in U.S. stock market history.

Meanwhile, OpenAI says the deep research tool is available to Pro users, with plans to roll it out to Plus, Team, and eventually Enterprise subscribers, marking a significant step towards the company’s broader goal of advancing towards artificial general intelligence (AGI).

Deep research is an agentic capability within ChatGPT designed to undertake multi-step research on the internet for complex tasks.