ChatGPT Deep Research

ChatGPT Deep Research is OpenAI's autonomous research agent that browses the web for 5 to 30 minutes to produce cited, comprehensive reports on complex topics. Rather than answering immediately, it plans a research strategy, browses dozens of sources autonomously, and synthesizes findings into a structured report with citations. The agent now runs on the GPT-5.5 model family (GPT-5.5 Instant, GPT-5.5 Thinking for advanced reasoning, and GPT-5.5 Pro for the highest accuracy), upgraded from the original GPT-5.2 launch. It scored 26.6% on Humanity's Last Exam (versus 9.1% for o1) and is state-of-the-art on the GAIA benchmark for autonomous web research. Pricing is tied to ChatGPT plan tiers. Free includes limited deep research queries. Go at $8/month provides more access to GPT-5.5 Instant with expanded messages and uploads but no expanded deep research. Plus at $20/month includes expanded deep research and agent mode with GPT-5.5 Thinking for advanced reasoning, projects, tasks, custom GPTs, and expanded Codex usage. Pro at $200/month provides the highest deep research limits with GPT-5.5 Pro and priority access. Business at $25/user/month billed annually includes everything in Plus with 60+ apps (Slack, Google Drive, SharePoint, GitHub, Atlassian), admin controls, SAML SSO, and no data training. Enterprise is custom-priced with enhanced security, compliance, and dedicated support. February 2026 added MCP client connectivity so users can connect Deep Research to any MCP server or app and restrict web searches to trusted sites, a meaningful enterprise unlock for teams that need research grounded in authenticated, industry-standard sources rather than the open web. July 2025 added visual browser integration via ChatGPT agent mode, expanding research capabilities to pages that require visual understanding. Users can track progress in real time and interrupt to refine with follow-up prompts or new sources. Deep Research is best suited for knowledge workers in finance, science, policy, legal, and engineering who need analyst-grade research outputs and can absorb the query caps on lower tiers. The agent handles extensive context and information gathering from diverse online sources, producing reports that would take a human researcher hours or days to compile. Deep Research is not appropriate for teams that need structured literature review of academic databases: Elicit and Consensus handle indexed paper databases (PubMed, arXiv, Semantic Scholar) better for systematic reviews of peer-reviewed sources. Teams needing daily high-volume research where Plus query caps become limiting face a steep upgrade to Pro at $200/month as the only viable tier. Q2 2026 state: GPT-5.5 model family now powers Deep Research, upgraded from the original GPT-5.2 launch. New Go tier at $8/month added between Free and Plus. MCP client connectivity confirmed for connecting to external data sources and restricting searches to trusted sites. Codex (AI-powered software engineering) now featured alongside chat as a core ChatGPT product. Visual browser via agent mode added July 2025. G2 shows 2,643 reviews at 4.6 for ChatGPT (parent product) as of June 2026. SOC 2 Type II, ISO 27001, and GDPR confirmed. Zero data training on Team, Business, and Enterprise tiers.