Consensus
- Consensus (consensus.app, Montreal/remote; founded 2021 by Eric Olson + Josh Kapla) = the citation-grounded academic answer engine — 250M+ peer-reviewed papers (Semantic Scholar + OpenAlex + PubMed + Crossref + publisher-licensed full text), natural-language yes/no questions → synthesized answer + Consensus Meter (agree/mixed/contradict ratio) + Study Snapshots + Deep Search agentic lit review. Not a general chatbot (ChatGPT/Claude), not a citation-context tool (Scite), not a systematic-review workbench (Elicit).
1. One-line positioning
Consensus = an AI research search engine that answers natural-language questions from 250M+ peer-reviewed papers using a hybrid retrieval stack (semantic + BM25 + relevance models) and GPT-5/Claude-routed synthesis, surfacing a Consensus Meter (what % of studies support/contradict), Study Snapshots (population/sample/method/outcome), Deep Search (agentic 50-paper lit review, 1,000-paper screen), Medical Mode (8M top-journal + 50k guideline subset), and citation export (APA/MLA/BibTeX/RIS) — Free $0 (3 Deep/mo, 15 Pro Analyses, 10 Snapshots) / Pro $15/mo ($120/yr) / Deep $65/mo ($540/yr) / Teams custom (~$12.99–20/seat) / Enterprise custom + Search API (~$0.10/req, MCP server for Claude/Codex/ChatGPT). It is the “what does the evidence say” tool, not the “write my essay” tool.
2. Core features
Layer Capability Notes Search NL yes/no + open questions, hybrid retrieval over 250M papers Abstract-grounded; full text when licensed or OA Consensus Meter Visual agree/mixed/contradict % on binary questions Signature feature; not in Perplexity/Elicit Study Snapshot Per-paper population/sample size/design/outcome pull Free 10/mo, paid unlimited Deep Search Agent expands query, screens up to 1,000 papers, 50 in synthesis, cites conflicts + gaps Pro 15/mo, Deep 200/mo Medical Mode Narrows to ~8M top-1,000-journal + 50k clinical guidelines AFib/CKD, tPA contraindications etc. Filters NL filters: “RCTs since 2020, adults, excludes animal” → auto-applied Study design/sample/population/year/journal tier Copilot / Ask Paper Follow-up Q on result set; “Ask Paper” on single DOI GPT-5 via OpenAI Responses API Export APA/MLA/Chicago/Harvard/BibTeX/RIS → Zotero/Mendeley/EndNote No PRISMA audit trail Surface 🌐 Web · MCP server (Claude/Cursor/ChatGPT/Codex) · Search API (enterprise) No mobile app, no desktop 3. Pricing (2026-08, reconciled across sources)
Tier Price Caps Free $0 Unlimited basic paper search, 15 Pro Analyses/mo, 3 Deep Searches/mo, 10 Study Snapshots/mo, 10 Ask Paper msgs/mo Premium / Pro $8.99–15/mo (annual $107.88–120/yr; monthly $11.99–15) Unlimited Pro Analyses + Snapshots + Ask Paper, 15 Deep Searches/mo, Consensus Meter, Copilot, export Deep $45–65/mo (annual $540/yr) 200 Deep Searches/mo, 50-paper synthesis depth, priority agent runs Teams $12.99–20/seat/mo (annual ~$240/seat) Central billing, up to 200 seats, +Deep allotment per seat, Search API access Enterprise / University Custom (library deals common; many students get Pro free via .edu) SSO, API ~$0.10/req, private corpus, 170+ university library partners - Conflicting numbers in wild: older pages show $8.99 Premium, newer show $15 Pro — encode both as “Premium $8.99–11.99 (legacy) → Pro $15 (2026 current), Deep $65”.
pricing_type = freemium;starting_price = $0;free_tier = "yes; 15 Pro Analyses + 3 Deep + 10 Snapshots/mo, unlimited basic search, no export batch";pricing_note = "Pro tier renamed Premium->Pro 2026 ($8.99 legacy vs $15 current); Deep $65/mo 200 Deep reviews; Teams $12.99-20/seat; uni library often grants free Pro; Search API ~$0.10/req enterprise; no mobile app"
4. Access Type (v2)
access_type: web-app, apiaccess_display: 🌐 Web app (consensus.app) · 🔌 MCP server + Search API (enterprise) · no iOS/Android/desktop5. Reviews (split)
Platform Score Volume Signal G2 4.7/5 ~1,550 Consensus Meter, time-savings, citation traceability Capterra 4.9/5 ~310 Intuitive vs PubMed, students love it Trustpilot 2.9/5 ~2–220 Billing/auto-renew/refund friction, abstract-oversimplification gripes Product Hunt / GetApp 4.9–5.0 — Early-adopter love Aggregate ~4.8/5 (G2+Capterra weighted) ~2,500 Trustpilot gap = billing pattern (same as Buffer/Canva/Character.AI/ChatGPT/ClickUp) Recurring praise: turns 3-hr PubMed scan into 5-min evidence read; Consensus Meter unique; great free tier for students. Recurring gripes: synthesis leans on abstracts (paywalled full text needs institutional link), not PRISMA-auditable for formal systematic reviews, Deep caps bite heavy users, Trustpilot billing complaints.
6. Best for / Not for
Best for Not for Grad students / postdocs scoping lit before deep dive Formal PRISMA systematic reviews (Elicit/Systematic Review Manager + manual screening) Clinicians answering “does X work” at point of care (Medical Mode) Live-web / market / policy-current events (Perplexity) Journalists fact-checking health/env claims Citation-context forensics (“how did later papers treat Smith 2019” → Scite) Policy analysts needing “state of evidence” fast Long-form essay drafting (Claude/ChatGPT) Researchers wanting MCP out to Claude/Codex workspace Teams needing offline/private-corpus indexing without enterprise deal University libraries (bulk Pro via institutional license) Pure paper-discovery/bibliometrics (Semantic Scholar/Connected Papers free) 7. Competitors (table)
Tool Lane Starts vs Consensus Elicit Screening/extraction/systematic review Free / $10–49/mo Deeper PRISMA workflow, extraction grids; weaker Consensus Meter, pricier Scite Smart Citations (support/contrast/mention) $20/mo Citation-context forensics, not question synthesis Perplexity Web-first cited answer $20/mo Live web/market/policy; not peer-reviewed-only, no Consensus Meter NotebookLM Source-pack synthesis (your PDFs) Free / $4.99 Synthesizes your corpus, not 250M-paper global answer Semantic Scholar Free paper discovery + TLDR $0 Broader free discovery, no GPT synthesis/Meter Research Rabbit / Connected Papers Citation-network mapping Free / $3–12 Field-mapping, not question-answering ChatGPT Deep Research / Claude General agentic research $20–200 Stronger reasoning, weaker paper-grounding + no Meter 8. Directory verdict (short_desc)
Consensus (consensus.app, 2021 Montreal; 250M+ papers, 170+ university library partners, 10M+ users) is the citation-grounded academic answer engine — NL yes/no questions → synthesized answer + Consensus Meter (agree/mixed/contradict %) + Study Snapshots + Deep Search (agentic 50-paper synthesis / 1,000-paper screen) + Medical Mode (8M top-journal + 50k guidelines) + hybrid retrieval (semantic+BM25+relevance) + GPT-5-routed synthesis + MCP server. Free $0 (15 Pro+3 Deep+10 Snapshots/mo) / Pro $15/mo ($120/yr; legacy Premium $8.99) / Deep $65/mo ($540/yr) / Teams $12.99–20/seat / Enterprise custom + Search API ~$0.10/req. G2 4.7 Capterra 4.9 Trustpilot 2.9 (billing). Category ai-productivity-data; tag academic-evidence-search; no new pipeline_stage. Distinct from Elicit (systematic workflow), Scite (citation stance), Perplexity (web), NotebookLM (your corpus), ChatGPT/Claude (general). Gotchas: abstract-grounded not full-text-audited, no PRISMA trail, Deep caps, Premium→Pro rename 2026, many .edu users get Pro free via library.