spaCy LLM is an open-source Python package from Explosion, the team behind the spaCy NLP library, that integrates large language models directly into structured NLP pipelines. Hosted on GitHub with about 1,400 stars, the MIT-licensed project is actively maintained, with the latest updates in 2026, and is distributed through PyPI for use in production NLP workflows.
The package adds a serializable llm component to a spaCy pipeline, so prompts become a normal, configurable pipeline step rather than ad-hoc API calls. Modular task functions define how a job is prompted and how the model’s unstructured response is parsed back into robust structured output, covering tasks such as named entity recognition, text classification, summarization and more, with no training data required. This spaCy LLM package is Explosion’s official extension for spaCy, and it is a distinct project from other similarly named LLM products listed elsewhere. Because it builds on spaCy’s configuration and training ecosystem, LLM steps can be combined with traditional statistical components in a single reproducible pipeline.
spaCy LLM is free and open source under the MIT license. You install it from PyPI at no cost; the only expenses are the API charges of whichever LLM provider you connect, or your own compute if you run open-source models locally.
NLP engineers and data scientists who already use spaCy and want to add LLM capabilities such as entity extraction or classification to existing pipelines without giving up structure, reproducibility or spaCy’s config-driven workflow.