Vectorize

Vectorize is Cloudflare’s vector database. You create an index, insert embeddings, and run similarity queries through a Workers binding. It is convenient and reasonably priced for adding semantic search or basic RAG to an existing Workers app.

The constraints appear quickly at scale. Index sizes are capped, metadata filtering is limited, and hybrid search is largely absent. Latency is acceptable rather than impressive. Using Workers AI for embeddings also ties Vectorize to that product’s reliability.

It is good enough for prototypes and modest production search features. Teams with serious retrieval workloads will eventually need a dedicated vector store. Treat Vectorize as convenient middleware, not a destination database.