Nougat
Meta's academic PDF parser that reads LaTeX maths and tables into Markdown.
Operated by Meta
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 a model from Meta's research group that reads a scientific paper and returns it as Markdown, including the mathematics as LaTeX and the tables as tables. It is aimed squarely at academic documents — it was trained on papers from arXiv and PubMed Central — and it is the open answer to the problem of getting equations out of a PDF as something a machine can use. It installs as a Python package, runs from a command line over a file or a directory, and can serve itself as a local HTTP endpoint.
Best for: Research groups and archives converting bodies of scientific literature into machine-readable text where the mathematics matters, and anyone building a corpus from arXiv or PubMed who needs LaTeX rather than a flattened approximation. It is one of very few tools that handles equations at all.
Worth knowing: The licence is split and the half that matters is not the code. The codebase is MIT, but the model weights are CC-BY-NC — non-commercial only — so an organisation that wants to use it in a commercial product has no licence to do so, and nothing in the repository offers one to buy. That is the opposite situation from Surya, where the same split exists but a commercial licence is available for sale. It is also narrow by training: the authors state it works best on English papers, that other Latin-based languages might work, and that Chinese, Russian and Japanese will not, which is a hard boundary for anyone with a mixed corpus. It needs a GPU to be practical, and the output is Markdown-compatible text rather than a PDF, so it is the front half of a pipeline rather than a document converter.
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 codebase; CC-BY-NC for the model weights, which permits non-commercial use only
- Redaction
- No redaction feature
- Programmatic access
- Embeddable library · Command line · HTTP API
- Last availability check
- Responded in 401 ms 16 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.