Weaviate
A self-hosted vector database that combines semantic and keyword search, reranking and RAG, with Docker, Kubernetes and managed cloud deployment.
Weaviate is a vector database for developers building search, recommendation and RAG applications. It stores data alongside embeddings so applications can retrieve related content and apply structured filters.
- Hybrid search: Combine semantic similarity with BM25 keyword matching and filters, or search images through the same database.
- RAG and reranking: Reorder retrieved results and connect them to an LLM for answers or summaries through a shared query interface.
- Embedding choice: Import existing vectors or generate them during import with integrated models, including OpenAI, Cohere, HuggingFace and Google.
- Shared deployments: Separate tenants, control access by role, and use replication and horizontal scaling for larger workloads.
- Memory and retention: Reduce memory use with vector compression and automatically remove stale objects with collection-level expiry rules.
The database runs locally with Docker, on Kubernetes, or in the managed Weaviate Cloud service. A local deployment can use a local embedding model; external model integrations send requests to the selected provider. Python, JavaScript/TypeScript, Java, Go and C#/.NET clients connect applications to the database, alongside REST, gRPC and GraphQL APIs.
Logotyper
