The short answer
Perplexity vs Elicit: pick Elicit, from Elicit, Inc., for academic and scientific research, because it searches more than 138 million papers, extracts data from many papers into one table and supports systematic reviews; its free Basic plan includes unlimited paper search and summaries. Pick Perplexity, from Perplexity AI, Inc., for everything else: business research, current news and quick fact checks with numbered citations from the open web.
| Perplexity | Elicit | |
|---|---|---|
| Made by | Perplexity AI, Inc., San Francisco | Elicit, Inc., United States |
| What it searches | The open web, plus Wiley journals and books on all plans | More than 138 million academic papers; the Research Agent also uses clinical trials, patents, regulatory and web sources |
| Free plan | Cited answers for limited daily usage (pricing page), basic models; Deep Research on paid plans | Basic: unlimited paper search, summaries and chat with full-text papers; limited Research Agent and reports |
| Citations | Numbered links on every answer | Sentence-level citations to papers and sources |
| Data extraction | Ask for tables from web sources | Extracts data from many papers into one table; Pro extracts up to 20 columns at a time |
| Systematic reviews | No dedicated workflow | Systematic Review workflow with PRISMA 2020 support; screen 5,000 papers on Pro |
| Agent | Perplexity Computer, paid with credits | Research Agent, the main way in since 26 August 2026 |
| Models | Choice of GPT-6.1 Sol, Claude Opus 5.5, Gemini 4 Argon and others on paid plans | Not named by Elicit |
| Paid plans | Pro, Max, Enterprise Pro, Enterprise Max | Pro, Scale, Enterprise |
| Training on your data | On by default for Free, Pro and Max until switched off; Enterprise never | Enterprise: no training by default; not stated on the pricing page for other plans |
The short verdict: Perplexity or Elicit?
If the answer to your question lives in peer-reviewed research, use Elicit. If it lives on the open web, in company pages, news, regulators or reports, use Perplexity. Elicit is a specialist: it searches more than 138 million academic papers and, since 26 August 2026, opens on its Research Agent, which gathers evidence from scientific literature, clinical trials, patents, regulatory databases, the web and your uploads, then produces cited reports, tables and slides. Perplexity is a generalist answer engine that searches the web for every question and, on paid plans, lets you choose the model doing the work.
Below is the winner for each kind of work. Everything is based on Elicit's and Perplexity's own pricing, changelog and help pages, not on tests we ran.
Research with sources: Perplexity wins for general research
For most working questions ("which CRM suits a 12-person agency?", "what changed in this law?", "who are the main players in this market?"), Perplexity is the right tool. It searches the whole web and answers in seconds with numbered citations, and Perplexity's pricing page describes its free plan as good for limited daily usage. Elicit is not a general chatbot: its sources are papers, trials, patents and regulatory records, so it is strongest in science and medicine and much less useful outside them.
Citations you can check: Elicit wins
Elicit's Research Agent produces reports with sentence-level citations, and its sources are papers, trials and regulatory records rather than whatever ranked on the web that day. When you click a citation you usually land on a study, not a blog post about a study. Perplexity's citations are always there, but their quality depends on what the web offers, so you often have to tell it to prefer official pages and reputable press.
Current news: Perplexity wins
Elicit is not a news tool. Perplexity searches the live web for every question, shows source dates when asked and, since 5 October 2026, can run scheduled digests with Automations in Perplexity Computer. Elicit has its own version of watching for new material: Routines, launched on 30 September 2026, monitor new evidence, update saved work and email summaries. That is useful for tracking new studies, not for daily news.
Academic papers: Elicit wins
This is Elicit's whole reason to exist. The free Basic plan gives unlimited search across more than 138 million papers, unlimited summaries and chat with full-text papers. Paid plans add the things that save real time: extracting data from many papers into one table, screening up to 5,000 papers on Pro (40,000 on Enterprise), reports drawing on up to 135 sources on Pro and 200 on Scale, and a Systematic Review workflow that supports PRISMA 2020 guidelines. Zotero import brings in your existing library.
Perplexity added Wiley journals and books on all plans in October 2026, which makes it better than it was for a quick scientific overview. But it has no screening, no extraction workflow and no systematic review support. For a literature review, a grant application or an evidence summary, Elicit is the clear winner.
Writing: Perplexity wins
Neither is a writing tool first. Perplexity wins because it is a general assistant with model choice: Pro lets you draft with GPT-6.1 Sol or Claude Sonnet 5.5, Max adds Claude Opus 5.5, and Perplexity Computer can produce documents and presentations. Elicit writes research reports well, but they follow a research structure; it will not help you with an email, a blog post or a pitch. If writing matters, pair either tool with a writing assistant such as Claude or ChatGPT.
Files and long documents: Perplexity wins for work files
For everyday business files, Perplexity is more useful: paid plans give more uploads, Projects keep files and connectors together, and @Microsoft 365 (Pro, Max, Enterprise Pro and Enterprise Max) searches SharePoint and OneDrive and reads Excel workbooks with a work or school account. Elicit's Research Agent also uses your uploaded data, and it is excellent with full-text papers, but it is built for research documents rather than contracts, decks and spreadsheets. If your files are papers, Elicit is better; for everything else, Perplexity.
Privacy: Perplexity wins
Perplexity documents its policy clearly. Free, Pro and Max train on your data by default, and you stop it by opening Account settings, then Preferences, and switching off AI data retention. Enterprise Pro and Enterprise Max are never used for training, uploaded files are kept for seven days, and Perplexity says outside model makers are contractually barred from training on its data.
Elicit's Enterprise plan has no training on your data by default, and Elicit says it has agreements preventing third-party providers such as OpenAI from training on Enterprise data. Its API does not use your inputs for training unless you opt in. But the pricing page does not state a training policy for Basic, Pro or Scale, and there is no documented user toggle. Perplexity wins because you can see the rule and change it yourself. If you handle unpublished or patient data, ask Elicit directly or use Enterprise.
Value: Elicit wins for researchers
Elicit's free Basic plan is unusually generous for its job: unlimited paper search, summaries and chat with full-text papers. The catch is that the Research Agent and reports, the parts that save the most time, are limited on Basic. Perplexity's free plan is the better free general research tool. If papers are any regular part of your work, Elicit's free plan is the best value on this page. Check the Elicit pricing page and the Perplexity pricing page for current prices.
Same task, both tools: does a four-day week improve productivity?
Task: "Our board wants to know whether a four-day working week improves productivity. Give me the evidence and a recommendation." This is how each tool handles it, based on documented features.
In Perplexity
- Ask what the evidence says, for and against, and ask for each source's year and who funded it. Because it searches the open web, expect the answer to mix trial reports, news coverage and studies, all cited.
- Ask for recent company and government pilots with dates. This is Perplexity's strength: news and reports sit outside Elicit's sources.
- Open the sources behind the key numbers before you quote them.
In Elicit
- Ask the Research Agent the same question. It searches the academic literature and other sources and returns a report with sentence-level citations.
- Ask it to extract, for each study, the country, sample size, sector, how productivity was measured and the result, into one table.
- On a paid plan, set a Routine to watch for new studies and email you when the evidence changes.
Perplexity tells you what the world is saying about the question. Elicit tells you what the research actually found and how strong it is. For a board paper, start in Elicit and use Perplexity for recent pilots and context.
Who should pick which
- Pick Elicit if you are a researcher, clinician, policy analyst, student or anyone whose decisions should rest on studies. It is not for news, business research or general questions.
- Pick Perplexity if your research is about markets, companies, products, laws or current events. It is not the right tool for a literature review or a systematic review.
- Teams that share research should note that Elicit's Scale plan adds real-time collaboration and an admin panel, and Enterprise adds SSO and SAML; on Perplexity, Projects act as shared workspaces and Enterprise Pro excludes your data from training.
- Use both if you write evidence-based reports: Elicit for the studies, Perplexity for the context. Elicit's MCP server on Pro or higher also connects it to Claude, ChatGPT and other MCP-compatible tools.
Learn both with our free lessons at /learn?tool=perplexity and /learn?tool=elicit, read How to use Perplexity and How to use Elicit, or see our pick of the best AI for research.
Prompts to copy
Evidence table (Elicit)
What does the research say about [question]? Find the most relevant studies and make a table with study, year, country, sample size, method, main result and limitations. Then summarise how strong the evidence is overall.
You get: A cited evidence table and an honest strength-of-evidence summary
Context and recent pilots (Perplexity)
What real-world trials, pilots or policy changes about [topic] have been reported in the last 2 years? List each with the organisation, country, date, result and a source. Use official reports and reputable press only.
You get: Dated, cited real-world examples to put next to the studies
Watch for new studies (Elicit Routine)
Every week, check for new studies on [topic] and update my [report name] report. Email me a short summary of anything that changes the conclusion.
You get: A recurring update when the evidence moves (paid plans)
Screen papers for a review (Elicit)
I am reviewing studies on [topic]. Find papers published since [year] that meet these criteria: [population], [intervention or exposure], [outcome], [study design]. List the included papers with a one-line reason each, and the excluded ones with the criterion they failed.
You get: A first-pass screen you can check before a systematic review