Back to Gap Map Infrastructure / Classic ML & computer vision

spaCy

open source / Overall score: 4.2(strong)

Python and Cython library for production natural language processing, running pipelines for tokenization, part-of-speech tagging, dependency parsing, named entity recognition, text classification, lemmatization and entity linking across more than 70 languages. It ships trained pipelines, a training system for custom ones and transformer support. Built by Explosion, which sells the separate Prodigy annotation tool and consulting.

Openness

5 high confidence

5.0

license: MIT(OSI)
source: public(github.com/explosion/spaCy)
core features withheld: no — Prodigy and consulting are sold as separate products

MIT, and the whole library builds from the public repository. Explosion describes spaCy as commercial open-source software and earns from Prodigy, a separately licensed annotation tool, and from consulting; neither is a withheld part of spaCy.

https://cdn.jsdelivr.net/gh/explosion/spaCy@master/LICENSE recorded 2026-09-26
LICENSE: "The MIT License (MIT)", "Copyright (C) 2016-2024 ExplosionAI GmbH, 2016 spaCy GmbH, 2015 Matthew Honnibal".
https://cdn.jsdelivr.net/gh/explosion/spaCy@master/README.md recorded 2026-09-26
README: "spaCy is commercial open-source software, released under the MIT license."; links to Tailored Solutions consulting.
https://data.jsdelivr.com/v1/packages/gh/explosion/spaCy@master?structure=flat recorded 2026-09-26
Flat file listing of master, 1,776 paths, with no ee, enterprise, commercial or pro directory.
https://prodi.gy/buy recorded 2026-09-26
Prodigy pricing: "$ 390 USD per lifetime license" for personal use and "$ 490 USD per seat, in packs of 5 seats" for companies - an annotation tool sold on its own license.

Adoption

5 high confidence

5.0

Measured on monthly PyPI downloads of the spacy package; the trained pipeline packages are not counted.

https://pypistats.org/api/packages/spacy/recent recorded 2026-09-26
last_month 21,026,566 downloads of spacy

Capability

3 medium confidence

3.0

spaCy is to classical NLP what scikit-learn is to tabular learning: a toolkit of trained components and the system to train more. Its pretrained pipelines handle a fixed set of linguistic tasks; a new label set still means training a component.

https://cdn.jsdelivr.net/gh/explosion/spaCy@master/README.md recorded 2026-09-26
README features: "Support for 70+ languages"; "Trained pipelines for different languages and tasks"; "Components for named entity recognition, part-of-speech-tagging, dependency parsing, sentence segmentation, text classification, lemmatization, morphological analysis, entity linking and more"; "Production-ready training system".

Verified 2026-09-26