Unpaywall is a free, open database that tracks the open-access (OA) status of scholarly articles and, where a legal free-to-read copy exists, links directly to it. It is built and maintained by OurResearch, a nonprofit organization, and is distributed as a browser extension, a REST API, and periodic full-database snapshots. For research administrators and librarians, Unpaywall functions less like a search engine and more like an infrastructure layer: other tools (library discovery systems, reference managers, institutional repositories, and bibliometric platforms such as OpenAlex) query it or its underlying data to answer one narrow, high-value question — is a version of this specific article legally free to read, and if so, where?
Who runs Unpaywall: OurResearch
Unpaywall is a product of OurResearch, a 501(c)(3) nonprofit founded in 2011 by Jason Priem and Heather Piwowar (originally operating as Impactstory before rebranding). OurResearch’s stated mission is to make scholarly research more open, discoverable, and reusable, and it builds its tools as open-source, open-data infrastructure rather than commercial products layered on closed data.
OurResearch’s best-known product beyond Unpaywall is OpenAlex, the free, CC0-licensed scholarly metadata catalog that succeeded Microsoft Academic Graph. The two products are closely linked at the infrastructure level: OurResearch has rewritten Unpaywall to run as a subroutine of the OpenAlex codebase, drawing on the same underlying data pipeline, and the organization has stated an intention to further align the open-access metadata the two products expose for the same DOIs so they don’t diverge. In practice, this means a research office already using OpenAlex for discovery or assessment work is drawing on infrastructure that overlaps substantially with Unpaywall, even though the two remain separate, purpose-built products — OpenAlex as a general scholarly-metadata catalog, Unpaywall as a narrowly scoped OA-status and full-text-location lookup. OurResearch also builds Unsub (subscription-value analysis for libraries, built directly on Unpaywall data) and continues to operate Impactstory.
How Unpaywall determines OA status
Unpaywall does not host content itself. It harvests OA signals from a large number of external sources — OurResearch has described monitoring tens of thousands of distinct content-hosting locations, including publisher websites, DOAJ-indexed gold OA journals, hybrid-journal OA articles, institutional repositories, and disciplinary repositories such as arXiv and PubMed Central — and reconciles those signals against a work’s DOI. Because publishers and repositories change access terms and post new copies continuously, Unpaywall’s underlying data is refreshed on an ongoing basis rather than compiled once; OurResearch has also introduced a public error-reporting flow (via unpaywall.org’s “fix” tool) so libraries and authors can flag incorrect OA determinations for correction, typically resolved within a few business days.
Unpaywall is distributed three ways:
- Browser extension (Chrome and Firefox). When a user lands on a paywalled article page, the extension checks the page’s DOI against Unpaywall’s database and, if a legal OA copy exists, displays an unlocked-padlock icon linking directly to it.
- REST API. A free, DOI-keyed lookup endpoint returns a structured JSON record for a given article, intended for integration into discovery layers, reference managers, link resolvers, and other institutional tools rather than for end-user browsing. OurResearch documents the API as requiring a contact email with each request (its access-control mechanism in place of an API key) and has published performance improvements following a 2025 rebuild onto the OpenAlex codebase.
- Full-database snapshots. Periodic bulk data downloads are available for organizations that need to work with the entire dataset offline — the model most bibliometric and discovery-tool integrations actually use, rather than millions of individual API calls.
The Unpaywall data model: OA status categories
Every article record Unpaywall returns includes an oa_status field with one of five values, plus a best_oa_location pointing to the specific free copy Unpaywall considers most authoritative when more than one exists:
- Gold — published in a fully open-access journal (typically one indexed in the Directory of Open Access Journals (DOAJ)).
- Hybrid — published in a subscription journal, but the specific article was made open under an explicit license, usually via an article processing charge.
- Green — published in a subscription (toll-access) journal, but a free copy is separately archived in an OA repository, such as an institutional repository or a disciplinary server like arXiv.
- Bronze — free to read on the publisher’s own site, but with no explicit open license and outside DOAJ-indexed gold OA journals; because there is no license, publishers can withdraw bronze access at their discretion. See CASRAI’s Bronze open access entry for more on this category, which Unpaywall’s own underlying research (Piwowar, Priem, Larivière et al., “The State of OA,” 2018) helped define and popularize.
- Closed — no legal free-to-read version could be located.
These categories are distinct from, but map onto, the color terminology used more broadly across scholarly publishing — see CASRAI’s Green open access and Gold open access entries, and the Gold vs Green vs Diamond vs Hybrid comparison for how OA route terminology is used across the field more generally (Unpaywall’s categories are an automated, article-level detection scheme built for a specific technical purpose, not a policy taxonomy).
Common research and library use cases
Collection development
Libraries use Unpaywall data (typically the bulk snapshot rather than live API calls) to identify which subscription journal content already has a legally free alternative available, informing decisions about which subscriptions still deliver unique access value. OurResearch’s own subscription-analysis product, Unsub, is built directly on this data for exactly this purpose.
OA-compliance checking
Research offices verifying compliance with funder or institutional OA mandates use Unpaywall (via API integration into repository or CRIS workflows, or via bulk data matched against a grant’s output list) to confirm whether a required deposit or OA route has actually resulted in a discoverable free copy, rather than relying on self-reported compliance alone.
Discovery
The browser extension is the most visible use case: individual researchers and students use it to find a free, legal copy of an article they’ve hit a paywall on, without resorting to informal file-sharing. Discovery-layer and link-resolver vendors also integrate the Unpaywall API so that a “free version available” indicator can surface directly in a library’s search results.
Unpaywall’s role in the open-access infrastructure ecosystem
Unpaywall is one of several tools that sit at the intersection of access, discovery, and authentication for scholarly content, each solving a related but distinct problem:
- Unpaywall answers whether a legal free copy of a specific article exists anywhere, and where.
- GetFTR (Get Full Text Research) is a publisher-backed service that tells a user, at the point of discovery, whether they have entitled access to the full text through their institution’s existing subscriptions, and routes them to it — a complementary rather than overlapping function, since it surfaces subscription access rather than OA copies.
- SeamlessAccess is a federated-authentication initiative that lets a researcher assert their institutional identity across publisher and platform sites without repeated logins, addressing the access-friction problem from the authentication side rather than the content-location side.
- DOAJ curates and vets which journals qualify as legitimate open-access venues in the first place — a quality and legitimacy layer that Unpaywall’s own gold-OA detection relies on, rather than a discovery or access tool itself. See CASRAI’s DOAJ vs Beall’s List vs Think.Check.Submit comparison for how these publisher-vetting tools relate to each other.
Together, these tools form a layered access stack: DOAJ vets which journals are legitimately open; Unpaywall detects, at the article level, whether a free legal copy exists anywhere (gold, green, hybrid, or bronze); GetFTR and SeamlessAccess handle the subscription-entitlement and authentication side for content that isn’t open. A research office building or evaluating a discovery layer, link resolver, or CRIS integration typically needs to understand where each tool’s responsibility starts and stops, since none of them alone solves the full “get the researcher to a legal full text” problem.
Frequently asked questions
Is Unpaywall free to use?
Yes. The browser extension, the REST API, and periodic full-database snapshots are all free. OurResearch operates as a nonprofit and funds this work through grants and paid services built on top of the same open data (such as Unsub), rather than by charging for access to Unpaywall itself.
Does Unpaywall host the articles it links to?
No. Unpaywall is a detection and indexing layer, not a repository. It links out to copies that already exist on publisher sites, institutional or disciplinary repositories, or other legitimate hosting locations — it does not store or redistribute full text itself.
How is Unpaywall different from OpenAlex?
OpenAlex is a general-purpose, CC0-licensed scholarly metadata catalog covering works, authors, institutions, and venues at large scale. Unpaywall is narrowly scoped to one question — OA status and full-text location for a given DOI. Both are built by OurResearch, and as of 2025 Unpaywall runs on the OpenAlex codebase, but they remain separate products with separate purposes; many institutions use both, often in the same workflow. See CASRAI’s guide to metadata search engines for scholarly research for how OpenAlex compares to other discovery tools.
Can Unpaywall data be wrong?
Yes, occasionally — OA status can change (a bronze article can be paywalled later, a green deposit can be added after initial publication) and automated harvesting from tens of thousands of sources will produce some errors. OurResearch provides a public correction mechanism at unpaywall.org for exactly this reason, and updates its detection logic periodically (a substantial rebuild in 2025 changed OA-status classification for a meaningful share of records).
