Elicit is an AI research assistant aimed squarely at people who read academic papers for a living, and its value depends heavily on whether your work involves the kind of structured, repetitive literature tasks it was built to automate.
What Elicit actually does
Most general search tools treat a research question as a bag of keywords. Elicit takes a different approach: it runs semantic search over a very large academic corpus, so a plain-language question surfaces relevant studies even when they do not share your exact terminology. The vendor states the platform draws on a database of over 125 million papers (its pricing page cites 138+ million), which is large enough that the practical constraint is rarely coverage and more often relevance filtering.
Beyond search, the tool is organized around a few concrete workflows. It can generate structured research reports that the company describes as inspired by the systematic review process. It extracts study-level details into a table format, so instead of reading twenty PDFs to note their sample sizes and outcomes, you get those fields laid out in columns. It supports a systematic literature review workflow that screens papers and pulls data, with Elicit reporting up to 80% time savings and search recall and abstract-screening accuracy figures in the mid-90s percent range. It also includes a library for organizing sources, alerts for newly published relevant work, and a chat interface for querying individual papers.
The reason these matter is specific to research practice. A systematic review or evidence scan is bottlenecked by two tedious phases: screening large candidate sets down to what is actually relevant, and extracting comparable data points from each included study. Both are mechanical enough to automate yet consequential enough that errors propagate into your conclusions. Elicit targets exactly these phases, which is why it lands better with evidence-synthesis teams than with someone doing a casual topic scan. You can find related options in the AI research assistant category.
Where it works in its favor
- Semantic search over a large corpus. Natural-language questions return conceptually relevant papers, which reduces the trial-and-error of keyword tuning.
- Structured extraction into tables. Pulling fields such as study characteristics and outcomes into comparable columns turns qualitative reading into something you can scan and sort. Paid tiers extend this to more columns and more data sources.
- Systematic-review screening at scale. The Pro tier screens up to 5,000 papers, Scale offers higher throughput, and Enterprise cites capacity for 40,000 papers and what the vendor calls PRISMA-grade accuracy.
- A genuinely usable free tier. The Basic plan includes unlimited search, unlimited paper summaries, unlimited chat with papers, and Zotero import, which is enough to evaluate the tool on real work before paying.
- Fits existing reference workflows. Zotero import is available on every plan, so the tool slots into a common academic setup rather than demanding you abandon it.
- API and collaboration on higher tiers. API access (Pro and above) and live team collaboration (Scale and above) support programmatic and multi-person work.
Where the caution flags are
- AI extraction still needs verification. High reported accuracy is not the same as perfect, and any extracted field or summary can be wrong. For publication-grade reviews you must check outputs against source papers; the time savings come from faster verification, not from skipping it.
- The corpus is a hard boundary. If a relevant study is not in the indexed database, the tool cannot surface it. Coverage of very recent, paywalled, non-English, or grey literature is not something the vendor documents in detail here.
- The heavier features are gated. The free plan explicitly limits Research Agent and Research Reports usage, so the workflows that distinguish Elicit from a plain search engine are the ones you have to pay for.
- Cost scales with ambition. Serious systematic-review throughput lives on the Pro tier and up, and the largest screening capacities and PRISMA-grade accuracy sit behind Enterprise custom pricing.
- It is domain-shaped. The interface and features assume an academic-paper mental model. If your evidence base is internal reports, patents, or web content rather than published literature, the fit weakens.
Pricing
Elicit uses a freemium model with three published paid tiers plus custom enterprise pricing. The figures below are from the vendor's pricing page.
| Plan | Price | Notable inclusions |
|---|---|---|
| Basic | Free | Unlimited search, summaries, and chat with papers; Zotero import; limited Research Agent and Research Reports usage |
| Pro | $49/month (billed annually, $588) | Systematic review screening up to 5,000 papers, 20 table columns, extraction from 135 sources, custom extractions, 10 research alerts, API access |
| Scale | $169/month (billed annually, $2,028) | 5x usage on core tools, figure extraction, live team collaboration, 30 table columns, 200 data sources, admin panel |
| Enterprise | Custom | Screening up to 40,000 papers, 40 table columns, PRISMA-grade accuracy, SSO/SAML/2FA, unlimited API, dedicated support |
The pricing tells a clear story: the free plan is for reading and asking questions, while the paid tiers are for structured screening and extraction at increasing scale. If your goal is a one-off scan, the free tier may carry you; if you run reviews regularly, the monthly cost has to be weighed against the analyst hours it displaces.
Who should consider it
Elicit fits academic researchers, PhD students, medical and clinical evidence teams, and policy analysts who repeatedly conduct literature reviews and evidence synthesis and want to spend less time on screening and manual data extraction. The reported time savings and screening capacities are most meaningful when you are handling large candidate sets rather than a handful of papers. It is a weaker fit for people whose sources sit outside the published-literature database, and for anyone unwilling to verify AI-generated outputs before relying on them. You can compare it against neighboring options in the tool directory or read related discussion on the blog.
Verdict
Treated for what it is, Elicit is a credible accelerator for literature-review work. Its strengths are concrete: semantic search over a very large corpus, structured extraction into comparable tables, and screening workflows that scale into the thousands of papers. Its limits are equally concrete: outputs need human verification, coverage is bounded by the indexed database, and the distinctive workflows are paid. If your research routine is built around systematic reviews or evidence synthesis, the free tier is worth testing on a real question, and the Pro tier is the point where the tool starts paying back the subscription in reclaimed time.
Common questions about Elicit
What data does Elicit search across?
The vendor states Elicit accesses a large academic-paper database, cited as over 125 million papers on the main site and 138+ million on the pricing page. Coverage of paywalled, non-English, or non-published sources is not documented in detail here.
Is there a free version, and what does it include?
Yes. The Basic (free) plan includes unlimited search, unlimited paper summaries, unlimited chat with papers, and Zotero import, with limited usage of the Research Agent and Research Reports features.
How much does the paid version cost?
Pro is listed at $49/month billed annually ($588), and Scale at $169/month billed annually ($2,028). Enterprise pricing is custom. Any monthly, non-annual rates are not published on the pricing page reviewed.
Can Elicit handle a full systematic review?
It supports systematic-review screening and data extraction, with capacity up to 5,000 papers on Pro and 40,000 on Enterprise, and the vendor cites PRISMA-grade accuracy on the Enterprise tier. Because AI extraction can contain errors, outputs should be verified against the source papers for publication-grade work.
Does Elicit fit into existing reference tools?
Zotero import is available on all plans, including the free tier. API access is included from the Pro tier upward for programmatic use.
