What is Deep Research?
TL;DR
An AI mode that parallel-reads dozens to hundreds of web sources, PDFs, and papers and produces a structured, cited report in 15-30 minutes — the dominant 2026 research feature, shipped by Gemini, Perplexity, and ChatGPT.
Deep Research: Definition & Explanation
Deep Research is the umbrella term for the AI feature class that emerged in 2024-2025 in which an agent autonomously decomposes a research question, parallel-browses dozens to hundreds of web sources, PDFs, and academic papers, cross-checks them, and produces a long-form structured report with citations. The big three: Google Gemini Deep Research (December 2024, Gemini Advanced / Workspace Business+), Perplexity Deep Research (February 2025, Pro $20), and OpenAI ChatGPT Deep Research (February 2025, Plus / Pro). Technically it combines (1) ReAct (think → search → read → evaluate loop), (2) multi-step planning (build the research plan first), (3) source triangulation (cross-check facts), (4) citation tracking, and (5) long-form synthesis (5,000-15,000-word structured outputs). Run time is 2-30 minutes; output approaches what a junior consultant would produce in a day. Pricing is $20-200/month, or pay-as-you-go via APIs (e.g., Perplexity Sonar API at $5 per 1M tokens). Common use cases include strategy and market research, technical and competitive analysis, academic literature review, investment due diligence, legal and regulatory research, and market intelligence. The primary risk is hallucinated citations — for high-stakes work, the human still needs to verify sources. By 2026, $20/month replaces a research analyst's worth of throughput, and adoption is now standard at consultancies, investment banks, and research labs.