Elicit
Consensus
The most complete scholarly research platform: systematic reviews built for PRISMA 2020, extraction tables, a Research Agent that reaches beyond papers, and API and MCP access
The fastest answer to what the research says, with a Consensus Meter and Deep Search at a low price
- PRISMA 2020 reviews
- Extraction tables
- Research Agent
- API and MCP
- Consensus Meter
- Free tier
- Deep Search
- Key features on higher tiers
- Unclear upper-tier pricing
- Preprints mixed in
- Less structure
Key Takeaways
- Elicit — 138 million papers and 500,000 clinical trials; PRISMA 2020 systematic reviews; Research Agent and Routines in 2026; API and MCP. 16/20.
- Consensus — Answers from papers with a Consensus Meter and Deep Search reports. Raised $30 million in May 2026. 15/20.
- Undermind — Deep search built for recall: reviews thousands of papers in about 2.5 minutes, monitors new work. Pro $16 a month billed annually. 14/20.
- Bottom line — Elicit for structured, auditable evidence. Consensus for quick answers on what studies show. Undermind to make sure you have not missed a paper.
How do Elicit, Consensus and Undermind differ?
All three are research agents that search scientific literature instead of the open web, so their citations point to studies. They split by job. Elicit structures evidence: it extracts data, writes reports and runs systematic reviews. Consensus answers questions: it tells you what the balance of studies shows. Undermind discovers: it searches until it has found the relevant papers on a hard topic.
Facts verified against vendor pages and coverage through October 31, 2026; scores match the Best AI Research Tools guide.
How do the features compare?
Feature matrix
| Feature | Elicit | Consensus | Undermind |
|---|---|---|---|
| What it searches | |||
| Corpus | 138M+ papers, 500K+ clinical trials; patents and safety reports via Research Agent | About 220M records, mainly Semantic Scholar and OpenAlex (third-party) | Scientific literature; open-access full texts plus your uploads |
| Full-text reading | Chat with full-text papers; extraction from figures on Scale | Summaries and study snapshots | Reads full texts and figures |
| Agents and workflow | |||
| Research agent | Research Agent (Aug 2026); Routines on a schedule (Sep 2026) | Deep Search reports | Deep search; topic monitoring |
| Systematic review | Guided workflow built for PRISMA 2020 | Not a systematic review tool | Discovery, not screening workflow |
| Agent access | API (Mar 2026) and MCP (Jul 2026) | MCP server on paid plans (third-party reported) | MCP; Add to ChatGPT and Claude |
| Buying (Oct 2026) | |||
| Free plan | Basic: unlimited search and summaries | Yes, with limited AI analyses | Yes, standard rate limits |
| Paid entry | Plus $11/user/mo billed annually | Not shown on pricing page; third-party lists Pro $15/mo | Pro $16/mo billed annually |
| WTF score (/20) | 16 | 15 | 14 |
Elicit
Elicit searches more than 138 million papers and 500,000 clinical trials. Its systematic review workflow supports PRISMA 2020 since May 2026. Research Agent arrived as a standalone product in August 2026 and Routines, which keep monitoring evidence, in September; an API and MCP server let other agents use it. Basic is free and Plus is $11 per user per month billed annually.
Elicit
- The only one of the three built for systematic reviews, with PRISMA 2020 support
- Extraction tables and exports for evidence work
- Research Agent reaches beyond papers to patents and safety reports
- Systematic review and larger reports sit on higher tiers
- Pricing page showed conflicting figures for Pro and Scale when checked
Consensus
Consensus summarizes what papers say with citations, shows a Consensus Meter for yes/no questions and writes Deep Search reports in a few minutes. It raised $30 million in May 2026 and says 2.5 million researchers start work with it each month. Its pricing page showed plan names but no prices when we checked.
Consensus
- Fastest way to see the balance of evidence on a yes/no question
- Cheap, with a usable free tier
- Strong funding and a large user base
- Index includes preprints; check venues before citing
- Less structure for formal reviews
Undermind
Undermind (Y Combinator, Summer 2024) runs a deep search that reviews thousands of papers, reads full texts and stops after about 2.5 minutes on average. It claims 73% recall against 29 to 39% for frontier-model agents on its own benchmark. Pro is $16 a month billed annually.
Undermind
- Built for recall: designed to find every relevant paper
- Monitors topics for new publications
- Low price and MCP access from ChatGPT and Claude
- Its recall advantage comes from a vendor-run benchmark
- Smaller company, seed-stage funding
Related questions
- Which AI research agents are worth using in 2026?
- Where do research agents fit in a company? (partner site CTAIO)
- Which general deep research tool should you use?
- What does Elicit cost and who is it for?
- How does Undermind search?
Frequently asked questions
Is Elicit or Consensus better?
They solve different problems. Elicit is better for structured evidence work: extraction tables, reports and systematic reviews built for PRISMA 2020. Consensus is better for a quick read on what studies say about a specific question, with its Consensus Meter. We score Elicit 16/20 and Consensus 15/20 on published capabilities.
What is Undermind best for?
Finding papers you would otherwise miss. Undermind searches iteratively through thousands of papers and claims much higher recall than frontier-model research agents on its own benchmark. It is a discovery tool to run at the start of a project, not a systematic review workflow.
Should I use ChatGPT deep research for literature reviews instead?
Use it for context, not the review itself. General research agents search the web, so citations often point to summaries and press coverage rather than the studies. Scholarly agents search paper databases directly, which makes each citation easier to check.
Are these tools free?
All three have free plans. Elicit Plus is $11 per user per month billed annually and Undermind Pro is $16 a month billed annually, as published in October 2026. Consensus did not show prices on its pricing page when we checked.
Final verdict
- Choose Elicit for literature and systematic reviews that need structured, auditable evidence. Elicit review.
- Choose Consensus for fast, cheap answers on what studies show. Consensus review.
- Choose Undermind to find every relevant paper before you start. Undermind review.
Sources
- Elicit, Blog and Pricing.
- Signalbase, Consensus funding (May 2026); Pasquale Pillitteri, Consensus AI guide 2026 (third-party).
- Undermind, Home and Pricing; Hunted/Product Hunt, Undermind (YC S24) launch (Aug 20, 2024); CB Insights, Undermind financials.
Ready to Find the Right AI Tools?
Browse our data-driven rankings to find the best AI tools for your team.