Industrial-strength open-source NLP library. By 2026, spaCy is widely used as a fast, deterministic complement to LLMs: handling tokenization, NER, and rule-bas
spaCy is industrial NLP for the pipeline era: fast, deterministic tokenization, NER, and parsing, now commonly paired with LLMs (spacy-llm) so structured extraction stays cheap and auditable while models handle the fuzzy parts.
Pricing: Free and open-source (MIT); Prodigy annotation tool sold separately.
Volume economics and auditability: at millions of documents, spaCy costs pennies and behaves identically every run. The 2026 pattern is hybrid: spaCy for structured backbone, LLMs for judgment calls.
Actively: including LLM-integration components. It owns the 'fast, deterministic NLP' niche that pure LLM stacks can't price-match.