spaCy

What is spaCy?

spaCy is an open-source Python library for natural language processing designed for real-world production use. Explosion has maintained it since 2015.

Key capabilities of spaCy

• Named entity recognition: Extracts people, places, organizations and more.
• Part-of-speech tagging: Labels each token with its grammatical role.
• Dependency parsing: Builds sentence structure trees.
• Text classification: Trains models to categorize text.
• Transformer integration: Supports BERT and RoBERTa pipelines.
• Visualizers and packaging: Built-in displays and model packaging tools.
• Custom components: Extend the pipeline with your own steps.

How spaCy works

You load a trained pipeline, pass text through it and read annotations such as entities, tags and parses from the resulting document object. spaCy is implemented in Cython for speed. Version 3 added reproducible training configs and a project system that takes work from prototype to production. spacy-llm adds prompting with large language models.

Who uses spaCy?

spaCy is used by Python developers and data scientists building information extraction, search and text analytics. Prodigy, from the same team, handles annotation.

spaCy pricing

spaCy is free and open source. Explosion sells separate paid products such as the Prodigy annotation tool and offers custom pipeline development services, but the library itself has no price.

spaCy alternatives

Alternatives include Hugging Face Transformers, which centers on pretrained transformer models, Google Cloud Natural Language, a managed API, and Amazon Comprehend, AWS managed text analysis.