turbopuffer

★ 4/5 · AI Search

Serverless vector and full-text search database built on object storage, giving engineers cheap, scalable retrieval for RAG and AI search apps.

Pros

Vector search database 10x cheaper than Pinecone Serverless with pricing recently reduced from $5/PB to $1/PB
handles 4+ trillion documents and 25k+ queries/sec in production
supports hybrid vector + BM25 full-text search

Cons

Less mature ecosystem and third-party integration library compared to Pinecone/Weaviate
potential vendor lock-in for large-scale deployments
costs scale linearly with data volume stored

Cost

Usage-based with plan minimums: Launch $16/month minimum; Scale $256/month minimum (audit log streams +$128/mo); Enterprise from $4,096/month with a 35% usage premium (single-tenancy, BYOC, 24/7 support, 99.95% uptime SLA); per-GB storage, write and query rates are shown only in the on-page cost calculator

Verdict

Best for startups and cost-conscious teams building vector search features who prioritize price efficiency over ecosystem maturity.

Visit turbopuffer