Nougat Base
Visual Transformer that converts scientific PDFs to markdown.
About
Nougat Base is Meta AI's 350M-parameter document understanding model, released August 2023 with the paper Neural Optical Understanding for Academic Documents. It is a Donut-style encoder-decoder: a Swin Transformer reads the full page image and an mBART-style decoder emits structured markdown, which makes it unusually good at math formulas and tables where line-by-line OCR engines fail.
Trained on over 8 million pages of scientific documents from arXiv and PubMed Central with their LaTeX sources, the base variant reaches roughly 89 BLEU on continuous text extraction in the authors' own tests (vendor-reported). Meta released the models and code openly; check the repository for current license terms before commercial use.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 133; Trending score 74/100 from researcher assessment.