All Tools
U
Vector DBFreeOpen Source
USEARCH
Single-file vector search with custom metrics and multi-language support
Apache 2.0
ABOUT
Implementing vector similarity search often requires either heavyweight database systems with significant operational overhead or limited libraries that only support standard distance metrics (Euclidean, cosine). USearch solves this by providing a single-file, embeddable vector search engine that supports arbitrary user-defined metrics, runs across 12+ programming languages, and maintains performance competitive with FAISS while being dramatically simpler to integrate. Its compact design makes it ideal for edge devices, mobile apps, and scenarios where a full vector database would be overkill.
INSTALL
pip install usearchINTEGRATION GUIDE
1. Add semantic search to a mobile or web application with minimal dependencies
2. Create a cross-modal search system using custom distance functions for text, images, and audio
3. Build a recommendation engine with domain-specific similarity metrics
4. Embed vector search directly into a game engine or IoT device
5. Prototype and benchmark approximate nearest neighbor algorithms before committing to a full DB
TAGS
vector-searchhnswnearest-neighborsimilarity-searchembeddingscpppythonjavascriptrustjava