Vectordbz

VectorDBZ is an open source desktop application for exploring, managing and analyzing vector databases. It runs on macOS, Windows and Linux and connects directly to the databases a team already uses, with no cloud proxy in between. Supported systems include Qdrant, Weaviate, Milvus, ChromaDB, Pinecone, pgvector on PostgreSQL, Elasticsearch and RedisSearch, with more listed. Users can keep multiple named connections with encrypted credential storage, browse collections, and run vector similarity searches with metadata filters, hybrid search and BM25 lexical search on databases that support it. A comparison view shows results side by side, and distance metrics include cosine similarity, Euclidean distance and dot product. For understanding embeddings, the app produces interactive 2D and 3D scatter plots using UMAP, t SNE or PCA for dimensionality reduction, with K means and DBSCAN clustering. Users can write custom JavaScript embedding functions, and templates are provided for OpenAI, Cohere and Hugging Face, plus a local test server. The tool suits developers and data engineers building retrieval and search systems who want a graphical client for inspecting what is stored in their vector index. The architecture is local first with HTTPS and TLS support, and the site states that data does not leave the user's machine. The software is free to download, with source on GitHub, and the site states there is no subscription.