Elicit prices its tiers around paper counts, not general chat

The writer's problem

A nonfiction writer or academic editor researching a claim often turns to a general chatbot and gets an answer with no way to check what it actually read. Elicit was built as a narrower tool: an AI research assistant that searches and summarizes published papers rather than answering from open-ended memory. Elicit's pricing page, as retrieved on 16 September 2026, states what the tool covers and what each paid tier adds, which matters to a writer deciding whether the product fits a literature-heavy chapter or a source-checking pass.

What the documents show

The pricing page states that the free Basic tier gives unlimited searches and summaries with limited usage for Research Agent and Research Reports, while the Elicit homepage describes the underlying index as covering over 138 million academic papers and 545,000 clinical trials. Paid tiers scale by task rather than by simple usage volume: Pro, at $49 a month or $588 a year, adds a systematic-review workflow screening up to 5,000 papers and extracting up to 20 data columns; Scale, at $169 a month or $2,028 a year, raises that to 30 columns and adds figure extraction and team collaboration; Enterprise, priced individually, is described on the same page as offering PRISMA-grade screening and extraction accuracy for reviews up to 40,000 papers. That PRISMA-grade phrase is Elicit's own characterization of its accuracy against a named systematic-review standard, not an independent audit, and the pricing page cites no outside evaluation to support it.

The editorial choice

This is an editorial checklist, not a recommendation: a writer citing a literature scan should distinguish between what Elicit's search returns and what its Research Agent has summarized, since the product's own materials separate a raw search from a synthesized report. An editor fact-checking a manuscript that leans on such a tool should ask which tier produced the material, since the free tier's usage caps differ from the customization Elicit describes for its paid Research Reports.

What stays with the author

Elicit's documentation describes what the software searches and extracts; it does not vouch for whether an extracted figure, once quoted in a manuscript, was read correctly in context. Verifying that a cited paper says what a table entry claims it says remains the writer's task, and the site does not claim otherwise.

• Did the material come from Elicit's search, its Research Agent, or a paid systematic-review workflow, and does the manuscript say which?
• Does a claimed PRISMA-grade accuracy figure come from an independent evaluation or only from the vendor's own description?
• Has every extracted data point been checked against the original paper before it appears in a draft?

A tool that defines its own scope this specifically is easier to check than one that claims to know everything; the specificity is the useful part, and it still leaves the checking to the person writing the book.