turbopuffer

turbopuffer is a vector and full-text search database built on object storage. It is for AI applications, semantic search, recommendation systems, and other workloads that need low-latency similarity search at scale.

It supports automatic scaling, hybrid search, metadata filtering, and full-text search. The product is designed for large production workloads, with support for billions of vectors, sub-10ms p50 latency, and very high write and query throughput.

Pricing is usage-based, and the product offers deployment options, multi-tenancy, security features like SSO, audit logs, IP allowlisting, private networking, and CMEK for higher plans.