Nougat
Meta Platforms' academic PDF parser that reads LaTeX maths and tables into Markdown.
Operated by Meta Platforms, Inc.
Opens the vendor's own site.
Jobs covered
Open a task to see exactly what the vendor documentation supports.
Extract
Browse all extract toolsextracting academic-document content with Nougat, where the README documents a visual encoder/decoder that reads a PDF and emits a .mmd Markdown file, mostly compatible with Mathpix Markdown, with LaTeX mathematics and tables. The CLI accepts one or more PDFs, a directory or a line-delimited PDF list, supports page ranges for a single PDF, selects the default 0.1.0-small or optional 0.1.0-base model tag, and writes one .mmd per PDF; it documents [MISSING_PAGE] failure output and the --no-skipping heuristic switch.
The scope records academic PDF transcription to Markdown/LaTeX/tables and the documented CLI selection, and makes no generic PDF object/text extraction, semantic validation, editing, redaction, conversion back to PDF, signing or repair claim
OCR
Browse all ocr toolsrunning neural OCR/transcription for academic PDFs with Nougat, where the paper describes a Visual Transformer performing an OCR task that processes scientific documents into a markup language, and the README documents raw page text with equations and tables emitted as per-document Markdown. The optional local API is nougat_api at http://127.0.0.1:8503/predict/, accepting a PDF and optional start/stop page parameters.
The scope records neural OCR/transcription of academic PDFs and the local API, and makes no general image OCR, broad-language, layout-understanding, guaranteed-accuracy, editing, redaction, conversion, signing or repair claim; English works best, other Latin-based languages might work, and Chinese/Russian/Japanese will not work
Where your files are processed
No server involved
A Python package and a model that run on your own hardware. The optional API mode is a local server on 127.0.0.1, so even that path keeps documents on the machine; the only fetch is the model weights. 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.
A Python package and a model that run on your own hardware. The optional API mode is a local server - the documentation gives it as http://127.0.0.1:8503/predict/ - so even that path keeps documents on the machine.
Model weights download from GitHub on first use, which is a fetch rather than a service.
What it does
Nougat is Meta Platforms' research model for academic-document transcription. A PDF is processed by a visual encoder/decoder and emitted as a .mmd Markdown file, mostly compatible with Mathpix Markdown, with LaTeX mathematics and tables.
It is intended for scientific papers like the arXiv/PMC material it was trained on. The package is installed as nougat-ocr: PyPI currently serves 0.1.17 (2023-10-04), the repository source declares 0.1.18, and GitHub's only releases are the 0.1.0-small and 0.1.0-base model weights (22 August 2023), not a software release.
The CLI accepts one or more PDFs, a directory or a line-delimited PDF list, supports page ranges for a single PDF and selects either model tag, saves one .mmd per PDF, and documents failure-detection output such as [MISSING_PAGE] and the --no-skipping option. An optional local HTTP API (nougat_api, http://127.0.0.1:8503/predict/) accepts a PDF and optional start/stop page parameters.
The code is MIT and the weights are CC-BY-NC; there is no commercial weights licence. It runs locally and requires the model weights, which may be downloaded on first use; it is not a hosted service.
Best for
Research groups and archives transcribing bodies of scientific literature into machine-readable Markdown where LaTeX mathematics and tables matter, using a local model rather than a hosted document service.
Worth knowing
The licences are split: the codebase is MIT, but the model weights are CC-BY-NC and non-commercial only. There is no commercial-weights option or paid licence in the repository, so a commercial product cannot rely on the weights.
The package and model versions are distinct: PyPI currently serves nougat-ocr 0.1.17 (2023-10-04), the repository source declares 0.1.18 (not a published package release), and GitHub releases 0.1.0-small and 0.1.0-base are model weights published on 22 August 2023, not a software release or launch date. The public repository does not expose a separate software release feed; do not describe a current software release beyond the package version.
Nougat works best on English scientific papers; other Latin-based languages might work, while Chinese, Russian, Japanese and other non-Latin writing systems will not work. It is trained on arXiv and PubMed Central-style papers, so it is not a general image OCR or broad-language OCR engine and output on other layouts is not guaranteed.
The output is Markdown/Mathpix-compatible text, not PDF, and it does not edit, redact, sign, encrypt, repair or convert back to PDF; there is no semantic validation guarantee. The documented [MISSING_PAGE] output and --no-skipping switch address a failure-detection heuristic, not a guarantee of page completeness.
Practical use generally needs a GPU and suitable PyTorch; the README notes CPU or older GPUs can trigger false positives, and model weights may download on first use. The optional API is a local nougat_api server at 127.0.0.1:8503/predict with start/stop page parameters, not a hosted service.
The separate open-source docTR library is unrelated. owner attribution to Meta Platforms, Inc. is supported by the repository LICENSE copyright and the project page, not merely the facebookresearch organisation.
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
- 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
-
MIT for the Nougat codebase; CC-BY-NC for the model weights, permitting non-commercial use only.
MIT for the Nougat codebase; CC-BY-NC for the model weights, permitting non-commercial use only. The repository LICENSE is MIT and its copyright line is Meta Platforms, Inc. and affiliates; the README states the code/weights split explicitly.
No commercial weights licence or paid tier is offered. The PyPI package is nougat-ocr and its current public release is 0.1.17 (uploaded 2023-10-04); the repository source declares 0.1.18, but that is not a published package release.
GitHub releases are model-weight releases 0.1.0-small and 0.1.0-base (22 August 2023), not a software version.
- Redaction
- No redaction feature
- Programmatic access
- Embeddable library · Command line · HTTP API
- Last availability check
- Responded in 440 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.
“This is the official repository for Nougat, the academic document PDF parser that understands LaTeX math and tables.”
What that does not establish
Recorded as `none` on absence: Nougat reads a PDF and emits Markdown, and nothing in its documentation describes removing content from a document. It does not write PDFs at all.
What it does not establish: that extracted text is safe. As with every extraction tool in this index, its output is the document's content - and its purpose is to make that content machine-readable, which is the opposite of redaction.
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.
-
“Nougat codebase is licensed under MIT. Nougat model weights are licensed under CC-BY-NC.”
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
- d5d1a548f65def9b22511bc14fdf2dbf6b77a2eedbf32e4890486e139827b404
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.
-
“Nougat was trained on scientific papers found on arXiv and PMC. Is the document you're processing similar to that? What language is the document in? Nougat works best with English papers, other Latin-based languages might work. Chinese, Russian, Japanese etc. will not work.”
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
- d5d1a548f65def9b22511bc14fdf2dbf6b77a2eedbf32e4890486e139827b404
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
- Which PDF tools keep your file on your machine - what we count as processing on your own device.
- Open source PDF tools, and what their licences require - what open source does and does not tell you.