Open-source vector database from Amsterdam with hosted embedding models and a Query Agent for natural-language queries.

turbopuffer AI features, pricing and GDPR (October 2026)
Serverless search engine on object storage for vector, full-text and hybrid search, with regions in North America, Europe and Asia Pacific.
from $16 per month
Pricing details
Usage based with minimum spend: launch from $16, scale from $256, enterprise from $4,096 per month (as of September 2026).
What is turbopuffer?
turbopuffer is a serverless search engine that keeps all data in object storage and caches it on NVMe SSD and memory for queries. It serves vector, full-text and hybrid search for AI applications with many namespaces and large document counts.
turbopuffer stores state only in object storage such as S3, which the vendor says keeps costs low, at the price of higher write latency and occasional slower queries on uncached data. It supports filters, full-text ranking, regex search, dense and sparse vectors and branching of namespaces. Regions run on AWS and Google Cloud in North America, South America, Europe including Frankfurt, Dublin and London, Asia Pacific including Mumbai, and Australia. Customers named on the homepage include Anthropic, Notion, Atlassian and Linear.
turbopuffer AI features and use cases
- Serves retrieval for AI products at scaleturbopuffer stores embeddings for many tenants and returns nearest neighbors combined with filters. Software vendors can offer search or RAG over each customer's data in separate namespaces.
- Combines vectors and keywordsBM25 full-text and vector search can be merged into hybrid queries. Results cover exact identifiers and semantic matches, which helps code and document search.
- Branches indexes for testingBecause object storage is the source of truth, namespaces can be branched as copy-on-write clones. Teams test new embeddings or chunking without copying the full dataset.
What sets turbopuffer apart
- Object storage as the only persistent state, with SSD and memory caching.
- EU regions in Frankfurt, Dublin and London on AWS, plus Google Cloud regions.
- Single-tenancy and BYOC on the enterprise tier.
Where turbopuffer is strongest
turbopuffer runs public clusters in North America, Europe, Asia, Australia and South America and does not state a country focus.
Questions to ask before you buy turbopuffer
Check whether higher write latency and occasional cold queries fit your application, as the vendor documents these tradeoffs. Choose the region explicitly for EU data. Single sign-on and audit logs start at the scale tier, and private networking and CMEK at enterprise. A SOC 2 report and a DPA are listed on all tiers.
Using turbopuffer in the EU: GDPR and AI Act
- Data locationMultiple regions, EU region available (according to the vendor). Hosting regions: United States, Canada, EU, United Kingdom, Australia, India, Brazil, Singapore, South Korea.
- Data processing agreementPublic, read it here (opens in a new tab). Subprocessor list (opens in a new tab).
- EU AI ActCheck whether the AI prepares decisions about people. If not, the high-risk rules usually do not apply. The high-risk obligations apply from December 2, 2027 (Digital Omnibus).
turbopuffer reviews
No reviews yet. Be the first to share how your team uses it.
Every review shows its verification level. We read each one before it goes live. How we check reviews
Questions about turbopuffer
What does turbopuffer cost?
Pricing is usage based with a monthly minimum: $16 on launch, $256 on scale and at least $4,096 on enterprise.
Can data stay in the EU?
Yes. turbopuffer lists AWS regions in Frankfurt, Dublin and London and Google Cloud regions in Europe. You choose the region when you connect.
Does turbopuffer support full-text search?
Yes. It supports BM25 full-text search, vector search and hybrid search that combines both.
turbopuffer update historyno updates yet
No changes recorded for turbopuffer yet.
Changes the vendor made and confirmed user reports, each checked against the vendor's own pages. We review every tool monthly.
Alternatives to turbopuffer
Tools in the same category (Embedding stores) with the most overlapping markets.
Managed vector database built on Milvus with hybrid search, a free tier, AWS, Google Cloud and Azure, plus bring-your-own-cloud.
Open-source embedding database with a four-function API, usable locally or as managed Chroma Cloud.
Open embedding model from the Beijing Academy of AI for more than 100 languages, with dense, sparse and multi-vector retrieval in one model.



