MIT Technology Review · Feb 3, 2025
MIT Technology Review reports OpenAI’s Deep Research launch
Markdown · JSON · Original source
MIT Technology Review reported on February 3, 2025, that had launched Deep Research, an agent for multistep online research. OpenAI says it can search and analyze text, images and PDFs, then produce reports with citations; users can also upload files. The publication describes the tool as powered by a version of OpenAI’s o3 , adapted for web browsing and data analysis. OpenAI presents the agent as a step toward and says a task taking it “tens of minutes” would take a human many hours. Those are OpenAI’s claims, not a demonstration of human-level performance in everyday use. The report also notes that Deep Research can facts and return incorrect information, though OpenAI says this happens at a notably lower rate than with ChatGPT. The tradeoff described is and compute: a query may take between five and 30 minutes, with longer research requiring more computing power. The report says the tool was available at no extra cost to Pro subscribers, with access for Plus, Team and Enterprise users planned. The evidence does not give the basis for OpenAI’s comparison with human analysts or its claim about hallucination rates.
Headline and summary by Escape Meridian, prepared by Luna from the attributed source. A source check is not independent confirmation of the source’s claims.
Original source
MIT Technology Review · Article
Source details
Original source date: Feb 3, 2025 (2025-02-03).
Date basis: Date from the publisher’s archive: its article address or archive listing.
Added to Escape Meridian: 2026-11-16T01:33:23.195Z. First observed: 2026-10-24T03:49:41.612Z. Retrieved: 2026-10-24T03:51:34.693Z.
Referenced by
Recorded citations and public file attachments. A reference alone does not establish agreement or influence.
No public references recorded here yet.
Limits
Text as retrieved on October 24, 2026; originally published February 3, 2025. The report gives no basis for ’s human-analyst comparison or its comparison of rates with ChatGPT.
Open question
What evidence supports ’s comparison of Deep Research with human analysts and its claim that it less often than ChatGPT?
Source-stated
MIT Technology Review reports that OpenAI launched Deep Research for multistep online research.
OpenAI has launched a new agent capable of conducting complex, multistep online research
Source-stated
The publication says Deep Research can search and analyze several kinds of material and compile reports with citations.
It can search and analyze massive quantities of text, images, and PDFs to compile a thoroughly researched report.
Source-stated
OpenAI says Deep Research works at the same level as a human analyst.
at what it claims is the same level as a human analyst.
Source-stated
OpenAI says Deep Research can sometimes hallucinate facts, but at a notably lower rate than ChatGPT.
OpenAI says the agent can sometimes hallucinate facts and present its users with incorrect information, albeit at a “notably” lower rate than ChatGPT.
Source-stated
The report says a Deep Research query can take five to 30 minutes and that longer research requires more computing power.
each question may take between five and 30 minutes for Deep Research to answer, it’s very compute intensive—the longer it takes to research a query, the more computing power required.
Source and contributor context
MIT Technology Review · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Reasoning model
A language model trained to work through a problem in written steps before it answers.
Artificial general intelligence
The idea of AI with broad problem-solving, creativity and adaptability across different tasks.
Hallucination
When an AI system states something false or made up as if it were true.
OpenAI
OpenAI develops AI models and products and publishes research, release announcements and evaluations.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: 03bb486d-3edd-5026-4850-84f2f4df48d8.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-16T01:28:45.865Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: 29dd55c35d8164a6c49f17e945bae3f7e738bf814b348e559340a2a280e02db7. Source text hash: 0c44f926fc8fe0db63fdc110f9108002c36d8c052ad9dab6df5251c866adc5da.
Revision and source-check history
Version 1 · 2026-11-16T01:33:23.195Z
MIT Technology Review reports OpenAI’s Deep Research launch
MIT Technology Review reported on February 3, 2025, that had launched Deep Research, an agent for multistep online research. OpenAI says it can search and analyze text, images and PDFs, then produce reports with citations; users can also upload files. The publication describes the tool as powered by a version of OpenAI’s o3 , adapted for web browsing and data analysis. OpenAI presents the agent as a step toward and says a task taking it “tens of minutes” would take a human many hours. Those are OpenAI’s claims, not a demonstration of human-level performance in everyday use. The report also notes that Deep Research can facts and return incorrect information, though OpenAI says this happens at a notably lower rate than with ChatGPT. The tradeoff described is and compute: a query may take between five and 30 minutes, with longer research requiring more computing power. The report says the tool was available at no extra cost to Pro subscribers, with access for Plus, Team and Enterprise users planned. The evidence does not give the basis for OpenAI’s comparison with human analysts or its claim about hallucination rates.
Original source
MIT Technology Review · Article
Source details
Original source date: Feb 3, 2025 (2025-02-03).
Date basis: Date from the publisher’s archive: its article address or archive listing.
Added to Escape Meridian: 2026-11-16T01:33:23.195Z. First observed: 2026-10-24T03:49:41.612Z. Retrieved: 2026-10-24T03:51:34.693Z.
Limits
Text as retrieved on October 24, 2026; originally published February 3, 2025. The report gives no basis for ’s human-analyst comparison or its comparison of rates with ChatGPT.
Open question
What evidence supports ’s comparison of Deep Research with human analysts and its claim that it less often than ChatGPT?
Source-stated
MIT Technology Review reports that OpenAI launched Deep Research for multistep online research.
OpenAI has launched a new agent capable of conducting complex, multistep online research
Source-stated
The publication says Deep Research can search and analyze several kinds of material and compile reports with citations.
It can search and analyze massive quantities of text, images, and PDFs to compile a thoroughly researched report.
Source-stated
OpenAI says Deep Research works at the same level as a human analyst.
at what it claims is the same level as a human analyst.
Source-stated
OpenAI says Deep Research can sometimes hallucinate facts, but at a notably lower rate than ChatGPT.
OpenAI says the agent can sometimes hallucinate facts and present its users with incorrect information, albeit at a “notably” lower rate than ChatGPT.
Source-stated
The report says a Deep Research query can take five to 30 minutes and that longer research requires more computing power.
each question may take between five and 30 minutes for Deep Research to answer, it’s very compute intensive—the longer it takes to research a query, the more computing power required.
Source and contributor context
MIT Technology Review · Publisher
The publisher is identified here. No individual author has been inferred.
Source media
Media retrieval status: unchecked.
Terms and references
Reference context selected by Escape Meridian; separate from source evidence. Entity mentions do not establish authorship, involvement or endorsement.
Reasoning model
A language model trained to work through a problem in written steps before it answers.
Artificial general intelligence
The idea of AI with broad problem-solving, creativity and adaptability across different tasks.
Hallucination
When an AI system states something false or made up as if it were true.
OpenAI
OpenAI develops AI models and products and publishes research, release announcements and evaluations.
TIME
TIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Editorial provenance
Version 1. Publication ID: 03bb486d-3edd-5026-4850-84f2f4df48d8.
Prepared by openai/gpt-6-luna; grounding review by anthropic/claude-opus-5.5.
Source check: passed by anthropic/claude-opus-5.5 on 2026-11-16T01:28:45.865Z. This checks fidelity to the source, not whether its claims are independently true.
Editorial selection: accept; reasons: bridge_admitted_by_model.
Observation: 29dd55c35d8164a6c49f17e945bae3f7e738bf814b348e559340a2a280e02db7. Source text hash: 0c44f926fc8fe0db63fdc110f9108002c36d8c052ad9dab6df5251c866adc5da.
passed · 2026-11-16T01:28:45.865Z · anthropic/claude-opus-5.5
Publication: 03bb486d-3edd-5026-4850-84f2f4df48d8
attribution — pass: The reading attributes the report to MIT Technology Review and the performance, AGI and hallucination-rate claims to OpenAI, as the source does. The page is one article, not a listing. Its 'Deep Dive' sidebar links are later page additions, and the reading does not use them. One small point: the reading frames 'search and analyze text, images and PDFs' as something OpenAI says, while the article states it in its own voice. This does not change the meaning.
OpenAI claims the tool represents a significant step toward its overarching goal of developing artificial general intelligence (AGI)
summary — pass: The reading leads with the actor and the development: OpenAI launched Deep Research, an agent for multistep online research. It explains what the tool does, what powers it, its claimed speed, its hallucination caveat, its compute tradeoff and its availability. The definitions list covers 'reasoning model', AGI and 'hallucinate', and the reading explains 'agent' in context. The title and question are neutral, and the question follows from the gaps the source leaves.
Deep Research will search the web, analyze the information it encounters, and compile a detailed report that cites its sources
date — pass: The reading gives February 3, 2025, matching originalPublished at day precision. The page was modified only on the same day. The planned rollout to Plus, Team and Enterprise users is reported as planned, not as an outcome. Sidebar items describe later events, such as the 'Ten years after AlphaGo' item, but they are later page additions and the reading does not use them. The kind is publisher-page, so no archive disclosure is required.
will soon roll out to its Plus, Team, and Enterprise users
support — pass: Every claim has a matching verbatim excerpt. The reading correctly treats the human-analyst comparison and the hallucination-rate claim as OpenAI assertions. It also correctly notes that the article gives no basis for either. The 5–30 minute and compute-intensity points are stated in the source. The reading infers no unsupported benefits or costs.
at what it claims is the same level as a human analyst.
media — pass: No media was selected, and no verified media candidates were supplied. Leaving media empty is appropriate.
Challenge or correct this reading
Send a correction or an opposing source with this reading’s URL. Disagreement and uncertainty can remain unresolved.
OpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIA language model trained to work through a problem in written steps before it answers.
Definition referenceThe idea of AI with broad problem-solving, creativity and adaptability across different tasks.
Definition referenceWhen an AI system states something false or made up as if it were true.
Definition referenceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEOpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIWhen an AI system states something false or made up as if it were true.
Definition referenceOpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIWhen an AI system states something false or made up as if it were true.
Definition referenceOpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIA language model trained to work through a problem in written steps before it answers.
Definition referenceThe idea of AI with broad problem-solving, creativity and adaptability across different tasks.
Definition referenceWhen an AI system states something false or made up as if it were true.
Definition referenceTIME is a news magazine publishing reporting on world affairs, business, technology and the workers behind digital services.
Explore TIMEOpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIWhen an AI system states something false or made up as if it were true.
Definition referenceOpenAI develops AI models and products and publishes research, release announcements and evaluations.
Explore OpenAIWhen an AI system states something false or made up as if it were true.
Definition reference