# Elicit — Research agent that searches 138M+ papers and runs systematic reviews with sentence-level citations.

> Source: The Agents Index — https://theagentsindex.com/elicit (structured, researched, re-verified)
> Facts last verified: 2026-09-14

Elicit is a research agent for scientists that searches academic papers, clinical trials, patents and regulatory databases, then produces cited reports, tables and systematic reviews. It runs screening and data extraction that researchers would otherwise do by hand.

| Fact | Value |
| --- | --- |
| Website | https://elicit.com |
| Price | from $75/month billed monthly, $49/month billed annually |
| Free tier | Yes |

Basic (free, standing): limited usage for Research Agent and Research Reports; unlimited paper search across 138M+ papers, unlimited summaries and unlimited chat with papers with full-text access; Zotero import. No Systematic Review workflow, no API or MCP access, no table columns/export tier, and Premium PDF parsing is unavailable on Basic. When the monthly usage pool is exhausted, Basic keeps Find Papers and Chat with Papers until the cycle resets.

## What it does

**Done-for-you service** · No — You run the software yourself on a self-serve subscription. Enterprise adds people for onboarding and training, but Elicit does not staff a team to build or operate research agents on your behalf.

> Dedicated customer success team providing training, onboarding, and success planning
> — https://elicit.com/pricing · Plan comparison cards, Enterprise column, sixth bullet

## How much it does unattended

**Runs autonomously** · Yes — Once a session starts the agent plans and executes end to end without supervision (you can close the tab and return to a finished output), though it may ask clarifying questions before it begins.

> The Research Agent breaks your prompt into a systematic plan, then executes it step by step. You can watch which sources it's consulting and how it's building toward your output in real time.
> — https://support.elicit.com/en/articles/14756886-elicit-s-research-agent · How it works, step 3 'Watch the agent work'

**Agent permissions** · Partly — The agent runs a whole plan unattended once launched; the only documented gates are the clarifying questions it asks up front, an effort slider, and an explicit approval before a new skill is saved.

> Nothing is saved until you approve it, and the options at that point are create, revise, or cancel.
> — https://support.elicit.com/en/articles/16647708-skills-in-elicit · Creating a skill, 'Ask the agent to build it'

_Not established: Multi-agent_

## Which models it runs on

**Claude** · Yes — Anthropic is a named subprocessor whose models generate Elicit's AI responses, so Claude is in the stack, but as Elicit's own choice, with no Claude toggle exposed to the buyer.

> Anthropic's models are used to give AI responses.
> — https://trust.elicit.com/subprocessors · Subprocessors table, Anthropic row, purpose column

**GPT** · Yes — OpenAI appears in the same subprocessor inventory with the same purpose, so GPT models also serve responses, again chosen by Elicit rather than selected by the user.

> OpenAI's models are used to give AI responses.
> — https://trust.elicit.com/subprocessors · Subprocessors table, OpenAI row, purpose column

**Open models** · No — Elicit's complete subprocessor inventory names exactly two model providers, both proprietary; the only other ML entry is Coreweave hosting Elicit's own custom models, and no open-weight option is offered to users.

> OpenAI's models are used to give AI responses.
> — https://trust.elicit.com/subprocessors · Subprocessors page, 'All subprocessors' complete 15-entry list — Anthropic and OpenAI are its only model-provider rows; no open-weight provider appears

**Model choice** · No — Elicit picks and ensembles models itself as part of its accuracy approach; no plan, settings page or API parameter anywhere on the site lets a buyer choose which model runs.

> We use strategies like process supervision, prompt engineering, ensembling multiple models, double-checking our results with custom models and internal evaluations
> — https://support.elicit.com/en/articles/14757979-elicit-s-reliability-does-elicit-run-the-same-risk-of-hallucination-as-other-ai-tools · Body text, third paragraph

## Where you use it

**In your editor** · Partly — Elicit ships no IDE or editor extension of its own; reaching it from an editor means connecting its hosted MCP server from any MCP-compatible client, which needs Pro or above.

> All three are also available through the Elicit MCP server, so Elicit can return cited answers inside Claude, ChatGPT, Gemini, or any MCP-compatible client.
> — https://support.elicit.com/en/articles/14757400-elicit-s-api · Body text, paragraph after the three endpoint bullets

**On the command line** · Partly — There is no first-party Elicit CLI product; the vendor publishes a CLI tool among its API integration examples, so terminal use means running that sample against the Pro-and-above API.

> Working examples in curl, Python, and JavaScript, plus integrations (CLI tool, Slack bot, Claude Code skill)
> — https://docs.elicit.com/ · Code Examples section, introductory sentence

**In your pipeline** · No — No CI or pull-request surface; Elicit's own surface enumeration is the web app and the API, and the closest published case is a customer's own scheduled jobs.

> The Research Agent is available through elicit.com or via API to bring into your workflows.
> — https://elicit.com/blog/introducing-elicit-research-agent · "Introducing Research Agent and BioDecisionBench" section

**In a browser** · Yes — The primary surface is the hosted web app at elicit.com, where sessions, projects, library and admin settings all live; the API is the alternative rather than the main route.

> The Research Agent is available through elicit.com or via API to bring into your workflows.
> — https://elicit.com/blog/introducing-elicit-research-agent · Introducing Research Agent and BioDecisionBench, first paragraph

## Whose machine it runs on

**Self-hosted** · Partly — Only the Enterprise column mentions custom deployments such as single-tenancy, and Elicit never documents that as running on customer infrastructure; every self-serve plan is vendor-hosted only.

> Enterprise-level security and controls, including SSO & SAML, 2FA, user analytics, domain verification, and custom deployments (e.g. single-tenancy)
> — https://elicit.com/pricing · Plan comparison cards, Enterprise column, fifth bullet

**Open source** · No — The terms reserve all rights in the service to Elicit and separately bar reverse engineering or decompiling it; only an API examples repo and a third-party licence list are public.

> Using the Service does not give you ownership of any intellectual property rights in or to any part of the Service.
> — https://elicit.com/operations/terms · Our Intellectual Property Rights in the Service and Feedback, third sentence

## What it costs to run

**How it meters** · Partly — Billing is per seat per month, but the work itself draws on a monthly usage pool whose size Elicit never states in any unit; overage is sold in fixed $100 increments of extra usage.

> Usage isn't a fixed amount per action, because the work varies from one task to the next.
> — https://support.elicit.com/en/articles/15646622-usage-limits-in-elicit · How usage is measured, fourth paragraph

**Free tier** · Yes — Basic is a standing free plan rather than a trial, giving unlimited paper search, summaries and chat with limited Research Agent and Report usage; no call or card is mentioned.

> Yes! While our most powerful workflows and features are on the paid plan, we also offer a generous free plan so you can use Elicit.
> — https://elicit.com/ · FAQ accordion, 'Is Elicit free?' answer

**API access** · Partly — A documented REST API covering search, reports and systematic review exists, but it starts at Pro; the docs list Basic and Plus as having no access at all to search requests.

> The Elicit API is available on the Pro, Scale, and Enterprise plans.
> — https://support.elicit.com/en/articles/14757400-elicit-s-api · Opening line, bolded

**MCP server** · Partly — Elicit runs a hosted MCP server with OAuth 2.0 so another agent can query it directly, exposing the full API surface, but the entitlement begins at Pro rather than the free plan.

> The Elicit MCP and API are available to customers on the Pro, Scale, and Enterprise plans.
> — https://support.elicit.com/en/articles/14757404-use-elicit-via-mcp-server · Opening line, bolded italic

**Bring your own key** · Partly — You bring your own documents and your institution's publisher subscriptions; the vendor documents no way to supply your own model keys or inference.

> Upload your own documents and integrate company subscriptions to access more full-text papers.
> — https://elicit.com/solutions/research-agent · Data sources section, BYOData card

## Buying it for a team

**A company can buy it** · Yes — An organisation buys a Scale team subscription billed centrally to an admin, or an Enterprise contract; invoice-based billing and institutional contracts are both documented separately.

> Centralized billing: All team members' subscriptions are billed to the Team Admin in a single purchase.
> — https://support.elicit.com/en/articles/14759330-elicit-s-scale-plan-for-teams · Scale includes list, 'Centralized billing' bullet

**Seat model** · Yes — Pricing is strictly per user per month with the buyer choosing seat count at signup and adding or removing seats later; no floor or ceiling is published, and removals get no prorated refund.

> When signing up, choose the number of team members on your subscription. You can add or remove team members at any point in your subscription.
> — https://support.elicit.com/en/articles/14759330-elicit-s-scale-plan-for-teams · Subscribing to an Elicit Scale plan section

**Pooled budget** · Yes — On a Scale team the included usage is the sum of the seats and everyone draws from one shared pool, so a heavy user can consume a quiet colleague's share without any per-seat split.

> A team's included usage is the sum of its seats, so a 4-member team has 4x an individual's usage, and the whole team draws from one pool.
> — https://support.elicit.com/en/articles/14759330-elicit-s-scale-plan-for-teams · Scale includes list, 'Pooled usage' bullet

**Admin controls** · Partly — Admin powers cover membership, seats, usage visibility and what happens to a departing member's work; nothing published lets an admin constrain what the agent may do or which sources it may reach.

> User management: Team admins have access to an Admin Dashboard to invite or remove team members, view usage, and purchase additional seats.
> — https://support.elicit.com/en/articles/14759330-elicit-s-scale-plan-for-teams · Scale includes list, 'User management' bullet

**Audit log** · Partly — Enterprise lists user analytics and Scale admins can view team usage, but no log of administrative actions is documented anywhere; the auditability Elicit advertises is of review decisions, not of admins.

> Enterprise-level security and controls, including SSO & SAML, 2FA, user analytics, domain verification, and custom deployments (e.g. single-tenancy)
> — https://elicit.com/pricing · Plan comparison cards, Enterprise column, fifth bullet

**Single sign-on** · Partly — SAML SSO with 2FA and domain verification appears only in the Enterprise column; OIDC is never mentioned, and self-serve plans up to Scale sign in with email or a Google social login.

> Enterprise-level security and controls, including SSO & SAML, 2FA, user analytics, domain verification, and custom deployments (e.g. single-tenancy)
> — https://elicit.com/pricing · Plan comparison cards, Enterprise column, fifth bullet

## What happens to your code

**Opt out of training** · Partly — Only Enterprise is promised no training on your data by default; no opt-out control is documented for Basic, Plus, Pro or Scale, though uploaded PDFs are stated to stay private to the account.

> No training on your data by default
> — https://elicit.com/pricing · Plan comparison cards, Enterprise column, third bullet

**Data residency** · Partly — Jurisdiction is disclosed and it is the United States only: every one of the fifteen listed subprocessors is US-based; no EU or UK region is offered, with transfers handled contractually by SCCs.

> Our servers are located in the United States.
> — https://elicit.com/operations/privacy · Is Your Information Transferred Internationally? section

**Getting out** · Partly — Data export is broad (citations to RIS and BIB, review stages to CSV or XLSX, reports to PDF or DOCX), but unused monthly usage expires each cycle and cancelling refunds nothing already paid.

> Unused usage does not roll over to the next month.
> — https://support.elicit.com/en/articles/15646622-usage-limits-in-elicit · When your usage resets section, closing line

**Certifications** · Yes — A Vanta trust centre publishes a SOC 2 Type 2 report plus SOC 3, Cyber Essentials, VPAT and HECVAT; GDPR and CCPA are handled through a DPA with standard contractual clauses.

> Elicit Research, PBC SOC 2 Type 2 Report 07152025.pdf
> — https://trust.elicit.com/ · Resources panel, SOC 2 section, report list

## Provenance

Every fact above comes from the vendor. Nothing here is independently corroborated yet.

## Coverage elsewhere

None recorded. Every fact above comes from the vendor and is not independently corroborated yet.

## What changed

- 2026-09-14 — Capabilities · Commercial terms · Free tier
- 2026-09-02 — Capabilities · Commercial terms · Free tier · Price
- 2026-08-31 — Capabilities · Commercial terms · Free tier · Price
- 2026-08-23 — Alternatives · Crunchbase link · LinkedIn link · One-liner · Summary

## Correcting this record

Anything marked "Not established" is a question we have not answered — it is not a No. If you represent Elicit, or anything here is wrong or out of date, you can supply it with a source: the operations this site accepts from an agent are listed at https://theagentsindex.com/.well-known/agents.json — that catalogue needs no key to read, and each entry states its own auth.

Supplying evidence is free. Nothing that can be bought here changes what a field says, whether a gap is closed, or how this record is scored.