Turbopuffer
A vector and full-text search database that stores data on object storage and bills by usage.
How the 5.8 was reached
- Quality of output5/10
Latency and scale figures are vendor-only; the independent evidence is a small benchmark repo with 48 stars. turbopuffer.com
- Ease of use7/10
Quickstart: create an API key, pip/npm install, upsert; minor gotchas like region and dimensions reported. turbopuffer.com
- Pricing value5/10
Launch plan has a $16/month minimum, Enterprise starts at $4,096/month, Scale is contact-only; no free tier listed. turbopuffer.com
- API and integration quality7/10
Docs cover vector, full-text and hybrid search plus an API overview; pip and npm clients. turbopuffer.com
- Solves the problem it claims to solve5/10
Claims of speed and low cost are supported mainly by vendor figures and a small benchmark repo. github.com
5.8/10 is the mean of the 5 criteria that apply. Scored on the five-criteria rubric v1, 29 September 2026. How scoring works
Pricing
- All database features
- Multi-tenancy
- SOC2 report, GDPR-ready DPA
- Community Slack and email support
- SSO and audit logs
- HIPAA-ready BAA
- Private Slack channel
- Support hours 8-5
- Single-tenancy and BYOC
- Private networking and CMEK per namespace
- 24/7 support and 99.95% uptime SLA
- 35% usage premium
Key Features
- Vector search, full-text search and hybrid search
- Namespace sharding up to 256TB in one index
- Python and npm clients
- Multi-tenant deployment on all plans
- Vendor cites 1T+ documents and 10M+ writes/s in production systems
Pros & Cons
Pros
- Docs cover vector, full-text and hybrid search plus chunking
- Quickstart is short: API key, install a client, upsert
- Every plan includes all database features
- Scales a single namespace up to 256TB, per the pricing page
Cons
- Speed and scale figures come from the vendor; independent evidence is a small 48-star benchmark repo
- No free tier is listed, and the Launch plan has a $16/month minimum
- Enterprise starts at $4,096/month with a 35% usage premium
- Vendor's own full-text p50 of 874ms is slow next to its 14ms vector figure
Suits teams that want a usage-billed search store for large vector or hybrid workloads and are willing to check the vendor's speed and cost claims themselves. Hobbyists who want a free tier should look at Chroma or Qdrant instead. The middling score reflects thin independent evidence, not a known flaw.
Try Turbopuffer →Added to scored.tools on
Competitors to Turbopuffer
Other tools in the database category worth comparing.
Pinecone
7.4/10Vector database for building knowledgeable AI applications with fast similarity search
Qdrant
8.2/10High-performance open-source vector database written in Rust for AI applications.
Weaviate
6.6/10Open-source vector database designed for AI applications and semantic search
Chroma
7.6/10Open-source vector database and search infrastructure for AI applications