open source vector database for embedding similarity search
Vearch is an open source vector database for similarity search over embeddings. It stores high dimensional vectors alongside ordinary scalar fields and returns nearest neighbours in milliseconds, so applications can search by meaning rather than by keyword. It is the engine behind large scale visual and semantic search, and it is commonly used as the memory and retrieval layer for retrieval augmented generation, recommendation and deduplication systems. Access is through a JSON REST API, with official Python, Go, Java and Rust clients and integrations for popular AI application frameworks.
This cloudimg image runs Vearch in its documented standalone mode, so a complete vector database is answering queries on a single machine as soon as it boots, with no cluster to assemble. The administrator credential is generated uniquely on your VM at first boot and is never shared between images, and the API is reachable only over TLS so that credential never crosses the network in clear text. Vector data and cluster metadata live on a dedicated data volume you can resize independently of the operating system disk, and the image ships with a walk through tested deployment guide and 24/7 support.