Skip to content
RankShark

turbopuffer

Serverless vector and full-text search on object storage.

ToolPaid · from $16/moby turbopuffersince 2023
67.1

SharkScore

#14 in Vector databases

Why it scores 67.1

turbopuffer has a SharkScore of 67.1/100, ranked #14 of 53 in Vector databases (updated 2026-11-19).

Each part compares turbopuffer with other listings in Vector databases. Evidence covers 87% of the score's weight.

Adoption
52 · 30%
User reviews
no data · 25%
Maintenance
100 · 20%
Trust & transparency
65 · 15%
Momentum
no data · 10%

Paying never changes this score. How ranking works · method v2026-11-18.2

RankShark review of turbopuffer

Automated evaluation by RankShark · 18 Nov 2026 · method v1.0 · How we review

We were not paid for this review.

As of 2026-11-18, turbopuffer is a vector and full-text search database built on object storage; its docs note higher write latency and occasional slow cold queries [E1][E3].

turbopuffer is a hosted search database for vector, full-text, and hybrid search. The vendor describes an architecture using object storage for state and NVMe SSD with memory caching for compute [E3].

Best for

  • Teams building vector, full-text, or hybrid search with metadata filters. · docs
  • Workloads that can trade higher write latency or occasional cold-query delays for an object-storage-based architecture. · docs

Not for

  • Workloads that require consistently low write latency or fast queries when data is uncached or unpinned; the vendor documents higher write latency and occasional slow cold queries. · docs

Key features

  • Vector, text, and hybrid search. The vendor says vector search can be combined with filters and supports dense and sparse vectors; it also lists text ranking and boosting and hybrid search. · home page, docs
  • Filtering and regex. The vendor says it supports filters against an inverted index and regex search using trigram indexes. · docs
  • Object-storage-based architecture. The vendor says object storage is the sole source of truth, with NVMe SSD and memory cache used for compute. · docs
  • Namespace branching. The vendor says branching creates copy-on-write namespace clones in constant time, with independent reads and writes afterward. · docs
  • Deployment choices. The vendor lists multi-tenant, single-tenant, and BYOC deployments. · docs

Pros

  • The docs list vector, full-text, hybrid, filter, and regex search features. · docs
  • The vendor offers multi-tenant, single-tenant, and BYOC deployment options. · pricing page, docs
  • The pricing page lists a $16/month minimum-usage Launch plan. · pricing page

Cons

  • The vendor documents higher write latency: p90 248 ms for 512 KB upserts. · docs
  • The vendor documents occasional cold queries, with p90 1,214 ms on 1 million documents for uncached or unpinned data. · docs
  • The Enterprise plan has a listed minimum of at least $4,096/month and a 35% usage premium. · pricing page
  • No free plan or free trial stated on the pages we checked.

Pricing

PlanPriceLimits
launch$16/month minimum usageIncludes all database features; multi-tenancy; community Slack and email support.
scale$256/month minimum usageIncludes all database features; multi-tenancy; HIPAA-ready BAA, SSO, audit logs, IP allowlisting, private Slack channel, and support hours 8-5. Audit Log Streams are +$128/mo.
enterprise>= $4,096/month; Contact us35% usage premium. Includes all database features; multi-tenancy, single-tenancy, and BYOC; private networking and CMEK; 24/7 support hours, support SLA, and 99.95% uptime SLA. Audit Log Streams are +$128/mo.

Checked 18 Nov 2026 · source

Free plan
Not stated (checked 18 Nov 2026)
Free trial
Not stated (checked 18 Nov 2026)
Trains on customer data
The vendor says it will not use Customer Data to train AI or machine-learning models or algorithms incorporated into the service for other customers without the customer's prior written consent. Its privacy policy also says personal information is never trained on or fed to a model used by another user. · privacy page, terms
Data retention
Not stated (checked 18 Nov 2026)
Data regions
Not stated (checked 18 Nov 2026)
Data processing agreement
The vendor says it provides a DPA for all customers. The terms say customers must enter into a DPA before including personal data in Customer Data. · terms, security page
AI models used
Not stated (checked 18 Nov 2026)
API
The docs list API sections for overview, write, query, namespace metadata, export, warm cache, list namespaces, delete namespace, and recall. · docs
SDKs
Not stated (checked 18 Nov 2026)
Webhooks
Not stated (checked 18 Nov 2026)
Platforms
Not stated (checked 18 Nov 2026)
Security claims
The vendor says customer data and usage data are encrypted in transit with TLS 1.2 or higher, and customer data is encrypted at rest with AES-256. Customer-managed keys are optional. · security pageThe vendor says it undergoes SOC 2 Type 2 audits of security and availability control design and operational effectiveness. · security pageThe vendor says customer-managed encryption keys are available on the Enterprise plan. · security pageThe vendor describes tenant-specific separation of data files and cache entries in its multi-tenant architecture. · security page

What we measured

Package downloads · 30 days
13,318,750
Tranco traffic rank · list of the day
#283,404
How we tested · sources
  • Hourly uptime and response-time checks of the product's URL.
  • Repository, release and download data from GitHub, npm, PyPI and Hugging Face where they exist.
  • We did not sign up, pay, or use the product as a customer.

Something wrong? Report an error with a public link that shows it; corrections re-run the review.

Top alternatives to turbopuffer

See all
  • 78.9

    Postgres platform with pgvector, auth and storage.

    ToolFreemium · from $25/mo
  • 75.3

    In-memory data store with vector search for AI apps.

    ToolFreemium · from $5/mo
  • 73.7

    Open-source search and analytics with a vector engine.

    ToolOpen source
  • 73.7

    Vector similarity search for Postgres.

    ToolOpen source