OpenAI has introduced a groundbreaking new ChatGPT feature called “deep research,” enabling AI to conduct in-depth web searches, compile reports, and provide sourced citations. Designed for intensive knowledge work, this tool marks a significant leap in AI-assisted research.

What Is OpenAI’s ‘Deep Research’ Agent?

In a Feb. 2 blog post, OpenAI announced the launch of deep research, a ChatGPT agent built for users in finance, science, policy, and engineering who require thorough, precise, and reliable research.

Unlike traditional AI responses, deep research actively browses the web, compiles findings into structured reports, and provides full citations. The process can take anywhere from 5 to 30 minutes, depending on the complexity of the query.

Beyond professional applications, OpenAI noted that the feature is also useful for consumer research, such as comparing big-ticket purchases like cars and appliances.

How Does Deep Research Compare to Other AI Models?

OpenAI claims that its deep research agent outperformed rival models on Humanity’s Last Exam, an AI evaluation consisting of 3,000 expert-level questions across 100+ topics.

• Deep Research Accuracy: 26.6%
• DeepSeek-R-1 (Chinese AI Model): 9.4%
• GPT-4o (OpenAI’s previous model): 3.3%

The deep research agent is powered by OpenAI’s latest “o3” reasoning model, optimized for web browsing and data analysis. It attempts to fact-check itself to minimize errors and misinformation.

However, OpenAI cautioned that deep research is not infallible—it can still hallucinate facts, make incorrect inferences, and struggle to distinguish authoritative sources from rumors.

The Competitive AI Landscape

The launch of deep research follows OpenAI’s recent unveiling of “Operator”, a ChatGPT agent capable of completing real-world tasks such as ordering groceries and booking vacations.

Meanwhile, Google has announced that its AI model Gemini will introduce a similar “Deep Research” feature in early 2025, intensifying the AI arms race.

Additionally, OpenAI and Microsoft are reportedly investigating whether DeepSeek, the Chinese AI firm that disrupted the market in January, improperly obtained ChatGPT’s API data—a sign of growing competitive tensions in the AI space.

Availability and Pricing

The deep research agent is currently available only to OpenAI’s $200-a-month Pro plan subscribers, with usage limited to 100 queries per month. This premium pricing suggests OpenAI is targeting professionals and enterprises rather than casual users, reinforcing AI’s growing role in high-level decision-making and analysis.

Conclusion

OpenAI’s deep research agent marks a significant advancement in AI-powered analysis, offering users a tool for fact-based, sourced research across complex topics. However, concerns about accuracy, misinformation, and competition from emerging AI rivals like DeepSeek suggest that the AI landscape remains volatile.

While OpenAI is marketing deep research as a productivity tool, its real power lies in data aggregation and influence. As AI agents shape what users see as “trusted” research, OpenAI is positioning itself not just as a knowledge provider, but as an information gatekeeper—controlling which sources and perspectives are prioritized in AI-generated reports.