Weaviate

Weaviate is an open-source vector database that stores data objects and text embeddings from machine learning models, enabling AI-powered similarity search and retrieval. Developed by SeMI Technologies (a Dutch company founded by Bob van Luijt and Etienne Dilocker in 2019, rebranded to Weaviate Inc.), Weaviate has over 11,800 stars as of 2026. Weaviate combines traditional object storage with vector embeddings, allowing developers to perform semantic search, similarity-based retrieval, and hybrid search combining keyword BM25 and vector search. Key features include: built-in vectorization modules supporting popular ML models (OpenAI text-embedding-ada-002 and text-embedding-3-small/large, Cohere, Hugging Face, Sentence-BERT, MiniLM, and multi-modal CLIP for text+image embeddings), or custom vectors provided externally, GraphQL API and REST API for data operations and queries, hybrid search combining BM25 keyword search with vector similarity (with adjustable alpha parameter to control the keyword-vector balance), nearVector and nearText search operators for similarity queries, multi-tenancy support (isolating data per tenant with on-demand tenant creation), vector index using HNSW (Hierarchical Navigable Small World graphs) for approximate nearest neighbor search with configurable ef_construction and max_connections parameters, automatic vector indexing on data ingestion, named vectors (storing multiple embeddings per object from different models), references (cross-references between objects for graph-like relationships), schema definition with classes and properties supporting data types (text, int, number, date, boolean, geoCoordinates, phoneNumber, blob, and object references), backup and restore (to S3, GCS, Azure Blob, and filesystem), role-based access control (RBAC) with custom roles and permissions, and multi-modal support (storing and searching text, images, and other media types).