nano-vectordb and vectordb
These are competitors offering lightweight, minimal-dependency vector database implementations in Python, with nano-vectordb prioritizing simplicity and local hackability while Jina's vectordb emphasizes a "just right" feature set for pragmatic production use.
About nano-vectordb
gusye1234/nano-vectordb
A simple, easy-to-hack Vector Database
Implements exact-match semantic search using flat-index nearest neighbor computation with numpy, persisting to JSON for reproducible indexing across sessions. Supports conditional filtering on arbitrary metadata fields and multi-tenant isolation via in-memory tenant management with configurable capacity limits, making it suitable for RAG prototypes and small-scale embedding workflows.
About vectordb
jina-ai/vectordb
A Python vector database you just need - no more, no less.
Leverages DocArray for vector search algorithms and Jina for scalable index serving, enabling CRUD operations with support for multiple backends (in-memory, HNSW). Deployable locally, as a microservice via gRPC/HTTP/WebSocket protocols, or on Jina AI Cloud with automatic sharding and replication capabilities.
Scores updated daily from GitHub, PyPI, and npm data. How scores work