ChatGPT Deep Research

ChatGPT Deep Research is the research mode built into ChatGPT. Users describe the outcome they need, ChatGPT proposes a research plan they can review and edit, and it then reads and analyzes many sources to produce a structured report with citations or source links. Users can follow progress, interrupt at any time to refine the focus and change which sources it can use. Reports open in a full-screen view with a table of contents, a sources used section and an activity history, and can be downloaded as Markdown, Word or PDF.

By default deep research can use the public web, uploaded files and connected apps. Depending on plan, region and workspace settings, sources can include document stores such as Google Drive and SharePoint and authenticated industry data services, and deep research uses only read actions from those apps. Research can be restricted to specific websites or set to prioritize them while still searching the wider web. It runs on OpenAI's latest models by default, and users can choose a legacy model instead.

Deep research is included in ChatGPT plans and priced in USD. Free is $0. Go is $8 per month in the US and may include ads. Plus is $20 per month billed monthly. Pro comes in three monthly plans, Pro 100 at $100, Pro 200 at $200 and Pro 500 at $500, and among individual plans only Pro 500 includes Ultrafast. New Pro 200 sign-ups, paused on September 10, 2026, have reopened. Business is $20 per user per month billed annually or $25 monthly for a Standard seat, and $100 annually or $125 monthly for a Premium seat with higher Work and Codex limits, starting at 2 users. Enterprise and Edu are priced through sales. Usage varies by plan: Free includes limited deep research, Pro includes the highest allowance, and an in-product counter shows remaining tasks, with fixed monthly allowances resetting every 30 days from first use.

Deep research follows the same data controls as regular ChatGPT conversations, so content on Free, Go, Plus and Pro can be used for model training unless the user opts out, while OpenAI does not train on Business or Enterprise data. Enterprise and Edu admins control access by role, and deep research activity is detailed in the Conversation API. HIPAA is available only under a Business Associate Agreement on eligible OpenAI healthcare and enterprise products, not on Plus or Pro. At its February 2, 2025 launch OpenAI reported 26.6% on Humanity's Last Exam for the model then powering deep research, a figure that describes the launch model rather than the models in use today, and noted that it can fail to convey uncertainty accurately.

ChatGPT Deep Research is best for ChatGPT users who want cited, multi-source reports on business, market, policy or technical questions without leaving ChatGPT. Researchers who need systematic reviews of peer-reviewed papers from indexed academic databases should consider Elicit, Consensus, ResearchRabbit or Semantic Scholar. Teams working in Google Workspace should consider Gemini Deep Research, and users who want fast answers with citations rather than a long report should consider Perplexity AI. Teams that need a report whose cited claims are checked against their sources before it ships, with a receipt to show for it, should consider Hyperresearch.

Current state Q4 2026: deep research runs on OpenAI's latest models by default with legacy models selectable, and availability depends on plan and country. Independent review evidence for this row is a disclosed proxy, since the G2 rating and review count shown belong to the ChatGPT product profile. OpenAI's trust portal, scoped to ChatGPT services including Enterprise and Edu, lists SOC 2 Type 2, ISO/IEC 27001 and ISO/IEC 42001, and OpenAI offers GDPR and CCPA data processing terms. ChatGPT Enterprise for FedRAMP is authorized through FedRAMP 20x, but OpenAI's feature list for FedRAMP workspaces does not include deep research, so no FedRAMP claim is recorded on this row.