An AI academic search engine over 220M+ peer-reviewed papers and preprints, designed to answer research questions with evidence-backed summaries rather than open-web results. Its standout feature is the Consensus Meter, which shows how the literature leans on a Yes/No question (how many papers say yes, possibly, mixed, or no), and Citation Grounding, which displays the exact quote behind every AI citation . It integrates directly into ChatGPT, Claude, and Microsoft 365 Copilot as an app, and offers an MCP server and REST API on paid plans . Pricing: Free (unlimited paper searches, 10 Pro messages/mo, 3 Deep reviews/mo, 30 API/MCP calls), Pro ($20/mo or $12/mo annual, unlimited Pro messages, 15 Deep reviews/mo, 500 API/MCP calls), Deep ($65/mo or $45/mo annual, 200 Deep reviews/mo, 2,000 API/MCP calls) . Teams and Enterprise custom. Used by 5M+ users; partners with 170+ university libraries .

Pros & Cons:

Pros: The Consensus Meter is a genuinely useful signal for Yes/No questions—it visualizes how the literature leans, which is faster than reading abstracts to gauge agreement; Citation Grounding shows the exact supporting quote behind each AI citation, directly addressing the hallucination concern that plagues general AI tools; the Free tier allows unlimited paper searches, making it accessible for exploration without payment; direct integration into ChatGPT, Claude, and Copilot means research happens inside the tool you’re already using, without tab-switching; the MCP server and API share one monthly call pool, and Pro includes 500 calls, which is workable for light programmatic use.

Cons: Independent evaluation flags limitations with complex queries—UBC’s librarian review noted oversimplified summaries and recommended pairing it with PubMed or Google Scholar for verification; no third-party security certification is published, and the privacy policy states it is not subject to CCPA, which matters for US-based institutional buyers; the corpus is English-first, with search ranking and summaries optimized for English queries; Deep review volume is capped by plan (3/mo Free, 15/mo Pro, 200/mo Deep), and API/MCP usage past the included allowance costs $0.05 per call; it searches academic literature only—not open web, news, or grey literature—so it is not a general-purpose research tool.

Who is it for?:Students, researchers, and clinicians who need to quickly find evidence and gauge consensus on a specific question; users already working inside ChatGPT, Claude, or Copilot who want academic search without leaving the conversation; teams that need a lightweight, lower-cost complement to Elicit (Consensus for quick evidence checks, Elicit for systematic reviews); institutions with library partnerships (170+ universities) that want full-text linking via their subscriptions. Not suited for users seeking open-web research or grey literature, teams requiring third-party security certifications (no public cert), or deep systematic review automation (Elicit’s PRISMA workflow is more purpose-built).