WASM-NATIVE VECTOR DATABASE
High-performance vector similarity search that runs entirely in the browser. HNSW indexing, binary quantization, and SIMD acceleration.
32x memory reduction with minimal recall loss. Store millions of vectors in browser memory.
WASM SIMD128 optimized distance calculations. 2x+ faster Hamming distance with @jsonMartin's contribution.
SQL-like query syntax for pre/post filtering. Combine vector search with structured queries.
No server required. Runs entirely in browser or Node.js. Zero external dependencies.
Hierarchical Navigable Small World graphs for logarithmic search complexity.
Save and load indexes. IndexedDB support for browser storage. Snapshot/restore capability.