OpenAI brings Deep Research to all European Pro users

If your research process usually involves opening 50 browser tabs and drowning in tabs you will never read, OpenAI wants to change that. Instead of generating instant, surface-level answers, ChatGPT is learning to sit down and do the heavy lifting for you.

OpenAI announced that its new Deep Research feature has officially rolled out to all Pro users across Europe. The expansion follows the initial debut of the autonomous research capability, bringing the agent to European subscribers.

Half an hour of thinking over instant answers

Deep Research is built around a specialized version of OpenAI's o3 model. It is designed to act more like a dedicated research assistant than a standard chatbot.

When given a complex prompt, the tool begins by asking clarifying questions to nail down what you need. It then heads off to search the web, reading and synthesizing information across multiple sources, including text, images, and PDF documents.

A single task takes anywhere from 5 to 30 minutes to complete. Once finished, users receive a notification and a comprehensive report complete with full citations.

The extended reasoning pays off on hard benchmarks. On Humanity's Last Exam, an evaluation designed to challenge advanced models, Deep Research scored 26.6%. For perspective, Gemini Thinking achieved 6.2%, while GPT-4o managed just 3.3%.

The $200 price tag and broader rollout

Right now, accessing Deep Research comes with a steep admission fee. It is restricted to ChatGPT Pro subscribers paying $200 a month, capped at 100 queries per month.

That limitation will not stay forever. OpenAI plans to expand access to Plus and Team tiers within weeks, provided safety metrics hold steady.

Completing the rollout to Pro subscribers across Europe signals that infrastructure and regional deployment are moving forward on schedule.

Why it matters

We think Deep Research marks a critical pivot for AI interfaces: moving from instant chatbots to autonomous agents that execute multi-step work.

For analysts, academics, and founders who spend days gathering market intel or reading whitepapers, getting a structured synthesis with citations in 30 minutes is genuinely useful. That said, at $200 a month, it is priced for business budgets. Most casual users should simply wait for it to land on regular Plus accounts.

Strategically, OpenAI is doing something smart. Instead of competing on token generation speeds, it is packaging long-chain reasoning and web browsing into a coherent workflow that directly challenges standalone AI search engines.

The catch is operational transparency. Spending up to 30 minutes waiting on a query means that if the agent misinterprets your prompt, you burn one of your 100 monthly queries on irrelevant data. Even so, Deep Research proves that agents taking hours of human labor down to minutes is no longer just theoretical.