
Weaviate
A vector database that embeds for you, searches by meaning and keyword at once, and isolates tenants
What is Weaviate?
Three things distinguish it from a bare index. Objects are stored with their properties, so a filter is part of the search rather than something applied afterwards; a vectoriser module can generate the embeddings on write, removing the separate pipeline that otherwise drifts out of sync; and multi-tenancy is a first-class construct where each tenant gets its own shard and can be parked or offloaded to object storage. It is a database you run and operate, though — for an embedded index inside one process, this is more machinery than the problem needs.
What can you do with Weaviate?
- Store objects, not bare vectors — A collection holds JSON objects with properties and a vector under a shared schema, so the metadata a query needs to filter on lives with the vector instead of in a separate lookup table.
- Let the database do the embedding — Vectoriser modules turn an object into a vector on write, so there is no second pipeline to keep in step with the data — and supplying your own vectors instead remains supported.
- Filter without losing results — The HNSW vector index is combined with an inverted index, which is what makes a filtered nearest-neighbour query stay high-recall rather than degrading into retrieve-then-discard.
- Give one object several vectors — Named vectors let the same object carry embeddings from different models, each with its own distance metric, index configuration and compression — useful when text and images need different treatment.
- Park a tenant instead of deleting it — Each tenant is its own shard with its own index, and tenants move between active, inactive and offloaded-to-cloud-storage states; Weaviate documents up to a million active tenants on modest cluster sizes.
Before you choose Weaviate
- This is a database process to run, back up and upgrade; if the requirement is an index embedded inside a single application, a library or a file-based store is the lighter fit.
- When the database generates the embeddings, the model becomes part of its configuration — changing to a different embedding model later means re-vectorising the collection rather than editing application code.
Star history
21 Aug to 30 Oct · +187
Frequently asked questions
Is Weaviate free for commercial use?
Weaviate is released under the BSD-3-Clause licence — OSI-approved open source, which permits commercial use.
How can Weaviate be deployed?
Weaviate is available as Self-hosted / Managed cloud.