Weaviate

Weaviate is an open-source AI-native vector database that differentiates itself through built-in vectorization modules, rather than requiring a separate embedding step before storing data, Weaviate integrates directly with embedding model providers (OpenAI, Cohere, Hugging Face) and generates vectors automatically at ingestion time. This integration means you can store raw text and Weaviate handles the embedding pipeline, simplifying the data ingestion architecture for teams that don't want to manage a separate embedding service. The GraphQL API provides flexible querying combining vector similarity with structured filtering in a single expressive query language. Weaviate's modular vectorizer architecture supports text, images, and multimodal content with different models handling each type.

Hybrid search combines BM25 keyword retrieval with dense vector similarity in configurable ratios, enabling retrieval quality improvements for queries where exact keyword matching complements semantic similarity. Generative Search modules pass retrieved documents to an LLM for response generation within the database query, enabling RAG patterns in a single API call. Deployment options include Weaviate Cloud (managed SaaS), Kubernetes via Helm charts, Docker for local development, and embedded mode for in-process use. The open-source core has an Apache 2.0 license.

Weaviate is most commonly adopted in enterprise search, knowledge base applications, e-commerce product search, and RAG systems where the built-in vectorization and GraphQL querying flexibility simplify the application architecture.