Full Text Search
Search by keyword or phrase with configurable tokenizers, stemming for 17 languages, and an extensible SQL-based query language.
Search by semantic meaning with the pre-installed pgvector extension and workflows that incrementally keep vectors in sync.
Surface results with higher precision and recall with hybrid search, which combines the strengths of full text and similarity search.
Full text search results are sorted by BM25, the term-based ranking algorithm used by ElasticSearch.
Bucket and collect statistical metrics on search results with faceted search.
ParadeDB automatically shards indexes, enabling developers to search hundreds of times faster compared to single-node Postgres.
Identify the most relevant portions of surfaced documents.
Build smart search boxes with autocomplete, which suggests documents most similar to a search query.
Tune your results by applying custom weights to individual fields.
ParadeDB's search engine is built on Tantivy, a Rust-based implementation of Apache Lucene.
An intuitive table editor makes it easy to explore your data and write queries.
ParadeDB is developed in public and actively welcomes community contributions.
Full text queries over one million rows return in under 100ms.