What OpenAI released

In its 2026-02-12 announcement, Introducing deep research, OpenAI positioned deep research as a new ChatGPT capability for handling complex, multi-hop analysis tasks. Instead of returning a single fast answer, the system is designed to browse multiple sources, compare evidence, and generate a structured report with citations.

How it is framed technically

OpenAI describes deep research as being built on an o3-family reasoning model optimized for browsing and synthesis. The practical shift is from prompt-response interaction to workflow execution: search, source triage, cross-checking, synthesis, and report writing in one loop. That architecture is relevant for market intelligence, policy tracking, and technical due diligence where users need both conclusions and inspectable references.

OpenAI also says rollout starts with ChatGPT Pro, Team, and Enterprise, with broader availability planned. This phased deployment suggests the company sees immediate demand in professional environments rather than only consumer chat use cases.

Benchmarks, caveats, and operations

For performance context, OpenAI cites 26.6% on Humanity’s Last Exam. The number indicates progress on difficult reasoning tasks, but it should not be read as a universal proxy for enterprise reliability. OpenAI explicitly notes limitations such as hallucinated details, citation formatting errors, and occasional source interpretation mistakes.

For teams, the important implication is operational. Deep research can reduce repetitive analyst effort, but only if organizations define review gates, source-quality policies, and approval workflows before decisions are made from generated reports. The feature’s value is strongest when combined with human verification rather than treated as a fully autonomous decision engine.

Overall, deep research represents a meaningful product step: ChatGPT is being pushed beyond conversational assistance toward evidence-oriented research orchestration, with enterprise governance requirements becoming part of normal deployment practice.

#o3

OpenAI Launches ChatGPT Deep Research for Multi-Step Web Analysis