Typesense
An open-source search engine combines typo-tolerant keyword and semantic search, with self-hosting, built-in RAG and optional managed cloud hosting.
Typesense is a self-hosted search engine for developers building site search, product catalogs or AI search over their own data. It combines keyword matching with searches based on meaning.
- Keyword search. Autocomplete, typo tolerance, synonyms and adjustable ranking help people find records as they type.
- Catalog discovery. Filters, facets, geographic search and dynamic sorting narrow results, while pinned records and grouped variants give catalog teams control over what appears.
- Semantic and vector search. Use your own embeddings or generate them with built-in S-BERT and E-5 models; hybrid search combines semantic results with keyword matches.
- Questions over your data. Built-in RAG answers questions using indexed content, and natural language search converts requests into filters, sorting rules and queries.
- Image and voice search. CLIP supports image searches by text or similarity, and Whisper transcribes spoken queries for search.
Typesense is open source under GPL-3.0 and can run on your own server or through Docker. Typesense Cloud provides managed hosting. Optional external embedding providers such as OpenAI process the queries and indexed data sent to them.
The engine keeps its search index in memory, so RAM needs grow with the indexed dataset. Scoped API keys restrict access to particular records, and replication supports high availability. Integrations include WordPress, WooCommerce, Laravel and InstantSearch.
