How we review: this assessment is based on the product’s official documentation, current pricing and published independent evaluations; it is not a hands-on test. Read our review methodology.
Most AI assistants will happily answer a research question; far fewer show exactly which papers support each sentence. Elicit was built for that second job. It began as a tool for finding and summarizing academic papers and, through a busy 2026 of releases — a Research Agent, Routines, team collaboration and an upgraded API — has become a broader platform for evidence synthesis. This review assesses Elicit as described in its official pricing and support documentation.
What it is
Elicit is an AI research assistant focused on scientific and academic literature. It searches a large corpus of papers, summarizes and answers questions about them with citations, extracts data into tables, and supports structured processes such as systematic reviews. It is aimed at researchers, analysts and organizations — Elicit says eight of the top 20 pharmaceutical companies are customers — rather than at casual web search. See our Elicit directory entry for a short summary.
Key features
- Paper search. About 138 million papers drawn from Semantic Scholar, OpenAlex and PubMed, deduplicated and updated weekly, plus a separate ClinicalTrials.gov corpus. Coverage includes journal articles, conference proceedings, preprints and working papers, but not books or dissertations.
- Research reports. Customizable briefs using a process Elicit describes as inspired by systematic reviews.
- Systematic reviews. A dedicated workflow for screening and data extraction, with screening quotes, exclusion reasons and, since May 2026, support for PRISMA 2020 guidelines.
- Data extraction. Custom extraction columns pull specific details — sample sizes, outcomes, methods — from many papers into a table.
- Research Agent. An agentic research environment, presented in its current form in August 2026, that can produce tables, figures and slides.
- Routines and Library. Routines keep work current as new research appears, and the Library stores sources that teams can share.
- API and MCP. Programmatic access and an MCP server for agents and other workflows.
Pricing
At the time of writing (October 2026), Elicit’s official pricing page lists:
| Plan | Price | Highlights |
|---|---|---|
| Basic | Free | Unlimited search, summaries and chat with papers; limited Research Agent and reports |
| Pro | $49/user/month, billed $588 annually | Systematic review workflow (5,000 papers), reports from up to 135 sources, 20 extraction columns, API access |
| Scale | $229/user/month, billed $2,748 annually | Five times Pro usage, reports from up to 200 sources, figure extraction, real-time collaboration, admin panel |
| Enterprise | Custom | Screening of 40,000 papers, SSO/SAML, no training on your data by default, dedicated support |
The pricing page offers monthly billing but did not display monthly prices; the figures above are annual rates.
Strengths
Transparency is Elicit’s core strength. Answers are tied to specific papers with sentence-level citations, and the support centre states plainly where the corpus comes from, how often it is updated and what it misses. That level of disclosure is rare among AI research tools and is exactly what researchers need to judge whether to trust a result. Our explainer on how AI research tools retrieve, rank and cite sources shows why this matters.
The free plan is useful in its own right. Unlimited search, summaries and chat with papers cover a large share of what students and occasional researchers need.
The systematic review workflow is serious. Screening with quotes and exclusion reasons, structured extraction and PRISMA 2020 support target real methodological requirements rather than generic summarization. Elicit cites customer case studies reporting large time savings; these are vendor-reported figures, but the feature set is aimed at the slowest parts of evidence synthesis.
Elicit is candid about its limitations. A dedicated support article explains that models can miss nuance or misread numbers, that results depend on the quality of underlying papers, and that users should actively look for evidence on several sides of a question.
Limitations
Price is the obvious barrier. The features that set Elicit apart — systematic reviews, larger reports, more extraction columns and API access — begin at $588 per user per year, and collaboration features require Scale at $2,748. For individual researchers without institutional funding, that is a significant cost.
Full-text coverage is partial. Elicit can analyse full text for open-access papers, or for subscription papers if you use its browser extension with your own journal access; otherwise it works only from titles and abstracts, which limits extraction accuracy.
Coverage has known gaps. Elicit itself notes that grey literature and some clinical trial reports are retrieved less reliably and that regional coverage varies, including gaps in Chinese publications. For exhaustive reviews it recommends supplementary searches in specialized databases.
Finally, Elicit cannot assess study quality. It will summarize a weak or flawed study as fluently as a strong one, so critical appraisal remains the researcher’s job. Our guide on how to read an AI research paper covers habits that apply to any field.
Who it’s for
- Academic researchers and PhD students working through large literatures.
- Systematic reviewers in medicine, public health and social science.
- Pharma, policy and consulting teams that need defensible evidence summaries.
- Students who want a free tool for finding and understanding papers.
Alternatives
- Perplexity — general web research with citations, cheaper but less specialized.
- Consensus — an academic search engine focused on summarizing findings.
- Semantic Scholar — a free academic search engine, and one of Elicit’s own data sources.
Verdict
Elicit earns 4 out of 5. Its documented strengths — a large, clearly sourced corpus, a generous free tier, sentence-level citations and a systematic review workflow built around real methodological standards — make it one of the most credible AI tools for academic research. We deduct a point because its most valuable features are expensive, full-text analysis depends on open access or your own subscriptions, and, as Elicit itself acknowledges, it cannot replace expert appraisal of the evidence.
