GROBID
Machine-learning extraction of scholarly PDFs into structured XML/TEI for citation and metadata mining.
Jobs Extract
Since 2011
Opens the vendor's own site.
Jobs covered
Open a task to see exactly what the vendor documentation supports.
Extract
Browse all extract toolsextracting and parsing structured data from PDF documents with GROBID, where the documentation describes it as a machine learning library that extracts, parses and restructures raw documents such as PDF into structured XML TEI encoded documents with a focus on technical and scientific publications, naming header extraction for title, abstract, authors, affiliations and keywords, references extraction and parsing, citation context recognition, full text extraction and structuring, PDF coordinates for bounding boxes and parsing of names, affiliations, addresses and dates, served through the REST endpoints /api/processHeaderDocument, /api/processFulltextDocument and /api/processReferences, which each take a PDF as a required multipart input and return TEI XML or BibTeX through the Accept header, with optional consolidation against Crossref or biblio-glutton and a documented 204 response when no content could be extracted, and with the error reference stating that a PDF with no text blocks, such as a scanned document without OCR, cannot be processed
Where your files are processed
Reported to stay on your device
Claimed by the vendor. Not yet independently verified. That is a property of the library itself: what an application built with it does with your file depends on that application, not on this.
Self-hosted is the documented deployment: your own Docker container or WAR, so the document is parsed on your own machine. Two things sit outside that, and neither is the whole PDF.
The project supports reference and funder consolidation through biblio-glutton and the CrossRef REST interface, with funder matching as an optional step; those paths may send extracted metadata to another service, which is metadata rather than necessarily the document. Separately the project runs a public demo server that accepts uploads and is rate limited and quota limited, and anything sent there has left your device.
Self-hosting is not by itself proof that nothing is sent anywhere, which is why both are named rather than folded into one answer.
What it does
GROBID reads a PDF and returns structured XML: a TEI-encoded document with the header metadata, the body, the bibliography and the in-text citations resolved as far as it can. It is built for scientific and technical articles rather than for general documents, and its accuracy on that shape of document is the reason it is widely embedded in scholarly search and repository pipelines.
It is a Java service, deployed as a Docker container or a WAR, and it only reads - it never writes a PDF back out.
Best for
Research groups, repository and discovery platforms, and anyone building a citation graph from a corpus of academic PDFs. It is built for that specific job, and it is the one here that emits TEI rather than plain text or Markdown.
Worth knowing
It is narrow by design. The models are trained on scholarly layouts, so a general business document, a form or a two-column newsletter is outside what it is tuned for, and the documentation does not present it as a general-purpose extractor.
It is also heavy to run: a Java service with deep-learning models, Docker, and enough memory that the project's own advice is to deploy your own instance rather than use the public demo, which carries quotas and exists for evaluation. The batch command-line mode is documented as deprecated, so a shell pipeline built on it has an expiry date.
Output is TEI XML, which is powerful and needs real post-processing before it is useful to anyone who is not already working in that format.
Details, and where each comes from
Vendor statements, our own reading of their pages, and the things nobody has established yet, each answered separately. Where we have not confirmed something it says so rather than guessing, and a row nobody has established is greyed out.
- Versions covered
- Self-hosted · Linux · Windows · macOS
- Website
- github.com
- Requires an account
- Not applicable
- Watermark on the free tier
- Not verified
- Maximum file size
- Not verified
- Batch processing
- Yes
- Data residency
- Self-hosted
- Open source
- Yes
- Self-hostable
- Yes
- Licence
- Apache-2.0 (documentation CC-0, annotated training data CC-BY)
- Redaction
- No redaction feature
- Programmatic access
- Embeddable library · HTTP API
- Last availability check
- Responded in 400 ms 21 October 2026
Checked automatically by our own runner, not reported by the vendor. 100% of our last 30 checks got a response.
What the redaction verdict rests on
From the vendor's own documentation, read 22 September 2026. Not tested by us.
“GROBID includes a comprehensive web service API, Docker images, batch processing, a JAVA API, a generic training and evaluation framework (precision, recall, etc., n-fold cross-evaluation), systematic end-to-end benchmarking on thousand documents and the semi-automatic generation of training data.”
What that does not establish
Recorded as `none` because the tool reads documents and emits XML - it does not write or modify a PDF at all, so there is no redaction feature and no output document in which content could be left behind. The basis is the project's own description of its surface, quoted above, and it establishes absence from the feature set rather than a test we ran.
What it does not establish: that every downstream system built on GROBID's output is safe, since the XML it emits contains the full extracted text.
What the vendor says
Claims are labelled rather than repeated as fact. Where we have checked something ourselves, it says so and shows when. Checked means we read the vendor's own page and it says what we recorded - not that we tested the behaviour, unless the note says we did.
-
“GROBID is a machine learning library for extracting, parsing and re-structuring raw documents such as PDF into structured XML/TEI encoded documents with a particular focus on technical and scientific publications.”
Checked by us - the source says this Read 1 October 2026
Confirmed. The quoted text appears in the rendered page as read on 2026-10-01.
This claim was carried as reported rather than checked until this read.
Standing: Not re-read against the live source
Check it yourself: the source, the SHA-256, and how we checked
- From
- the text a browser renders
- SHA-256
- d5c45ca97e4bc8ff8bef931a147e99187d73e53f1365be4329f0535396ce70c4
How we checked
Read as rendered text in headless Chrome on 2026-10-01 and saved to the evidence archive, so the read can be repeated against the same file. Comparison ignores case, punctuation and hyphenation, which is why "open-source" on the record matches "open source" on the page.
-
“First developments started in 2008 as a hobby, following a suggestion by Laurent Romary (Inria, France). In 2011, the tool has been made available in open source.”
Checked by us - the source says this Read 1 October 2026
Confirmed. The quoted text appears in the rendered page as read on 2026-10-01.
This claim was carried as reported rather than checked until this read.
Standing: Not re-read against the live source
Check it yourself: the source, the SHA-256, and how we checked
- From
- the text a browser renders
- SHA-256
- d5c45ca97e4bc8ff8bef931a147e99187d73e53f1365be4329f0535396ce70c4
How we checked
Read as rendered text in headless Chrome on 2026-10-01 and saved to the evidence archive, so the read can be repeated against the same file. Comparison ignores case, punctuation and hyphenation, which is why "open-source" on the record matches "open source" on the page.
Each claim is quoted from the vendor's own pages. This section describes what we have and have not checked - it is not a judgement about anyone.
Sponsorship is distribution, not an endorsement or a promise of search rankings, AI citations, traffic, or sales. This listing is a free entry, compiled from published sources.
Related reading
- Open source PDF tools, and what their licences require - what open source does and does not tell you.