OpenAI Responds to DeepSeek Hype with ‘Deep Research’ ChatGPT Agent

Expert system company OpenAI has actually released Deep Research study, a brand-new AI-powered representative within ChatGPT that separately performs thorough web research study, examines information, and assembles reports– finishing jobs that would take human beings hours or perhaps days.

The tool, offered now for Pro users, is created for specialists in financing, science, policy, and engineering, along with anybody searching for “comprehensive, exact, and trustworthy research study,” according to OpenAI’s Feb. 2 statement.

The release comes at a time of growing competitors in AI research study, with China-based DeepSeek turning the script on AI economics, declaring it can match ChatGPT’s intelligence without the heavyweight price.

OpenAI CEO Sam Altman hyped the tool on Twitter, calling it “like a superpower; specialists as needed!”

Altman shared an individual usage case where Deep Research study assisted him discover an unusual NSX cars and truck in Japan after he had actually almost quit browsing by hand.

Users can access Deep Research study by picking it in ChatGPT’s message author, inputting an inquiry, and connecting files or spreadsheets for extra context.

The AI then examines numerous sources over 5 to thirty minutes, showing a sidebar tracking its research study procedure before providing a completely mentioned report.

Regardless of its sophisticated abilities, Deep Research study isn’t without its peculiarities.

OpenAI yields that the design “can often hallucinate realities or make inaccurate reasonings” and has problem with separating reliable sources from false information.

Presently, Pro users can access 100 questions each month, with Plus and Group users anticipated to get gain access to next, followed by Business clients.

OpenAI is supposedly dealing with a much faster, more cost-efficient variation to increase question limitations for paid users.

To reinforce its case, OpenAI indicated Deep Research study’s efficiency on Humankind’s Last Test, an AI criteria screening expert-level understanding throughout 100+ fields.

Deep Research study’s o3 design scored 26.6% precision, far exceeding DeepSeek-R-1’s 9.4% and even OpenAI’s GPT-4o, which handled just 3.3%.

The timing of OpenAI’s launch is no coincidence– last December, Google presented its own “Deep Research study” function for its Gemini AI design for Gemini Advanced users.

The AI giants are now secured a fight to specify the future of AI-driven, self-governing understanding discovery– a far cry from the chatbot period that began all of it.