turbopuffer
Limited infoServerless vector search built on object storage for low cost.
Limited info: public information on turbopuffer is limited, so this is a starting point. Verify current details with the vendor.
turbopuffer is a managed, serverless vector and full-text search engine built on object storage, aimed at large datasets at a fraction of in-memory cost. It targets high write throughput and pay-for-use economics.
License
Proprietary
Deployment
Managed
Pricing
Usage-based on storage and queries.
SDKs / languages
Python, TypeScript
Founded
2023
Strengths
- Very cheap at large scale (object storage)
- Serverless, pay-per-use
- Combines vector + full-text
Limitations
- Proprietary managed only
- Higher cold-latency than in-memory engines
Alternatives
Chroma
Developer-friendly open-source embedding database.
LanceDB
Embedded, serverless vector database on the Lance columnar format.
Milvus / Zilliz
Scalable open-source vector database for billion-scale search.
MongoDB Atlas Vector Search
Vector search built into MongoDB Atlas documents.
pgvector
Vector similarity search as a Postgres extension.
Pinecone
Fully managed, serverless vector database.
Facts verified 19 July 2026. Neutral summary, not an endorsement; verify current details with the vendor.