Typesense is built for instant search experiences, like the results that appear as a user types in a store or documentation site. It keeps its index in memory for speed, handles typos by default, and supports filtering, faceting, sorting, and grouping. Its API is simple, and client libraries exist for popular languages.
It also supports modern retrieval needs. Vector search lets applications find results by meaning using embeddings, and hybrid search combines that with keyword matching. Built-in features can generate embeddings, and a conversational search mode answers questions using retrieved documents. Developers can run Typesense on their own servers or use Typesense Cloud.
Typesense is free to self-host under an open-source license, with paid hosting priced by resources rather than by search volume. Because the index lives in memory, large datasets need enough RAM. Alternatives include Algolia, Meilisearch, Elasticsearch, OpenSearch, and Pinecone for vector search. Typesense suits developers who want fast, easy search they can control.