altor-vec
A vector similarity search engine written in Rust that compiles to a 54 KB WebAssembly module, letting a web application run semantic search entirely in the browser with sub-millisecond query latency and no server, API key or per-query cost.
altor-vec implements HNSW-based approximate nearest-neighbor search compiled to WebAssembly, so a browser can search on the order of 10,000 vectors in under a millisecond without sending queries to a server. It targets cases such as a documentation site wanting semantic search without a paid hosted service, an application wanting typo- and meaning-tolerant search without a server, or a privacy requirement that queries never leave the user's device.
Unlike text-similarity libraries such as Fuse.js, which match characters, altor-vec matches on meaning through vector embeddings, so a query like 'cancel plan' can match content about ending a subscription; this requires embeddings to be generated ahead of time and bundled as a search index. The project publishes a comparison of its bundle size and latency against Algolia, Fuse.js, Orama and Voy, and provides quickstart guides for React, Next.js and Vue.
It installs from npm and is licensed under MIT.