Building RAG applications or similarity search? Stop wrestling with heavyweight vector databases that need their own infrastructure. Zvec is what SQLite is to traditional databases - a zero-config, embeddable vector store that runs directly in your process. Drop it into any application and start querying millions of vectors with sub-millisecond latency.
Built on Alibaba’s battle-tested Proxima engine (the same tech powering their e-commerce search), Zvec handles both dense and sparse vectors, supports hybrid search with filters, and runs everywhere from your laptop to production servers. The API is refreshingly simple - install via pip, load your vectors, and search. No Docker containers, no cluster management, no YAML configs. With 5.3k stars and growing fast, it’s becoming the go-to choice for developers who want enterprise-grade vector search without enterprise complexity.
Perfect for prototyping ML applications, building recommendation engines, or adding semantic search to existing apps. The Python bindings make it trivial to integrate with your ML pipeline, while the C++ core ensures you’re not sacrificing performance for convenience.
⭐ Stars: 5386
💻 Language: C++
🔗 Repository: alibaba/zvec