Deep Research is an
AI agent
Agentic AI is a class of artificial intelligence that focuses on autonomous systems that can make decisions and perform tasks without human intervention. The independent systems automatically respond to conditions, to produce process results. The ...
integrated into
ChatGPT
ChatGPT is a generative artificial intelligence chatbot developed by OpenAI and released on November 30, 2022. It uses large language models (LLMs) such as GPT-4o as well as other Multimodal learning, multimodal models to create human-like re ...
,
which generates cited reports on a user-specified topic by autonomously browsing the web for 5 to 30 minutes.
Agent
Deep Research can interpret and analyze text, images, and
PDF
Portable document format (PDF), standardized as ISO 32000, is a file format developed by Adobe Inc., Adobe in 1992 to present documents, including text formatting and images, in a manner independent of application software, computer hardware, ...
s.
It is based on a specialized version of OpenAI's
o3 model.
Deep Research scored 26.6% on the "
Humanity's Last Exam
Humanity's Last Exam (HLE) is a language model benchmark consisting of 2,500 questions across a broad range of subjects. It was created jointly by the Center for AI Safety and Scale AI.
Creation
Stanford HAI's AI Index 2025 Annual Report cites ...
"
benchmark
Benchmark may refer to:
Business and economics
* Benchmarking, evaluating performance within organizations
* Benchmark price
* Benchmark (crude oil), oil-specific practices
Science and technology
* Experimental benchmarking, the act of defining a ...
, outperforming rivals like
DeepSeek's model
R1 (9.4%) and
GPT-4o
GPT-4o ("o" for "omni") is a multilingual, multimodal generative pre-trained transformer developed by OpenAI and released in May 2024. It can process and generate text, images and audio. GPT-4o is free, but ChatGPT Plus subscribers have higher ...
(3.3%).
According to OpenAI, Deep Research occasionally makes factual
hallucinations
A hallucination is a perception in the absence of an external stimulus that has the compelling sense of reality. They are distinguishable from several related phenomena, such as dreaming ( REM sleep), which does not involve wakefulness; pse ...
(errors) or incorrect inferences.
It may also reference
rumors
A rumor (American English), or rumour (British English; American and British English spelling differences#-our, -or, see spelling differences; derived from Latin 'noise'), is an unverified piece of information circulating among people, especial ...
,
and may not accurately convey uncertainty.
On April 24th 2025, OpenAI announced that a 'lightweight' version of Deep Research was to be released to quench the demand placed on the company by users.
Deep Research is currently offered to ChatGPT Pro subscribers ($200/month), who receive 250 queries per month (150 of which being 'lightweight'), to Plus, Team and Enterprise users who receive 25 queries per month (15 of which being 'lightweight'), and to Free users, who receive 5 'lightweight' queries per month.
References
External links
*
{{OpenAI
2025 in artificial intelligence
ChatGPT