Softwr

Databases · head to head

turbopuffer vs Zilliz

turbopuffer logo

turbopuffer

Databases

Closed-source vector and full-text search service built directly on object storage, with cold queries measured in seconds rather than milliseconds.

From
$16/month
Rated
-
Full reviewWebsite
Zilliz logo

Zilliz

Databases

Managed vector database and vector lakebase for AI applications

From
Free
Rated
-
Full reviewWebsite

The short version

  • Only Zilliz has a free tier, so it costs nothing to try first.
  • Each has a real cost: turbopuffer a cold namespace pays object storage latency on the first query, with a documented p90 around 1,214 ms on a million documents, so any interactive search box needs the data kept warm or the user waits about a second.; Zilliz pricing structure not publicly disclosed, requires sales contact
  • They diverge on capability: turbopuffer covers Object storage architecture, Zilliz covers Vector indexing.
  • Prices and features above were last checked on 30 August 2026.

Where they differ

Only the attributes on which turbopuffer and Zilliz actually diverge.

Attributes where turbopuffer and Zilliz differ
AttributeturbopufferZilliz
Starting price$16/monthFree
Pricing modelsubscriptioncontact-sales
Free tierNoYes
PlatformsWebCloud, Self-hosted
FoundedUnknown2017

Identical on both: user rating (Not yet rated), category (Databases).

What each one covers

Drawn from each product's published feature list. An absence here means we hold no record of it - not that the product lacks it.

Only in turbopuffer

  • Object storage architecture
  • Namespaces
  • Vector search
  • Full-text search
  • Attribute filtering
  • Documented limits
  • Configurable consistency
  • Durable writes

Only in Zilliz

  • Vector indexing
  • Distributed architecture
  • SQL interface
  • Tensor support
  • Real-time search
  • Cloud-native
  • Open-source compatible

What people use each for

The jobs each tool is most often brought in to do.

turbopuffer

  • A product with one search index per customer and thousands of customers, most of whose data is idle on any given daynot Zilliz
  • Very large corpora where holding every vector in memory is the dominant cost and occasional cold-query latency is acceptablenot Zilliz
  • Hybrid retrieval combining BM25 and vector search where running and synchronising two separate systems is the problem being solvednot Zilliz
  • Retrieval for agent and assistant products where indexes are created and destroyed frequently and per-index overhead must be near zeronot Zilliz

Zilliz

  • Build retrieval-augmented generation (RAG) systemsnot turbopuffer
  • Implement semantic search over documentsnot turbopuffer
  • Create multimodal search with text and imagesnot turbopuffer
  • Power recommendation engines with vector similaritynot turbopuffer
  • Enable similarity search on user embeddingsnot turbopuffer

Where each one falls short

Documented limitations, not opinions. Every one is a constraint you would hit in normal use.

turbopuffer

  • A cold namespace pays object storage latency on the first query, with a documented p90 around 1,214 ms on a million documents, so any interactive search box needs the data kept warm or the user waits about a second.
  • Queries are eventually consistent by default, and after roughly 128 MiB of outstanding writes new data is invisible until indexed, which the vendor puts at tens of seconds for small namespaces and tens of minutes for large ones, so a bulk re-index is not immediately queryable.
  • It is closed source with no community edition, so single-tenant or bring-your-own-cloud deployment is a commercial negotiation rather than a deployment choice, and there is no path to running it yourself if the relationship ends.
  • Per-namespace ceilings, roughly 10,000 writes per second, 32 MB/s and 500 million documents per shard, mean a single enormous index has to be sharded across namespaces by your application rather than by the service.
  • It is a search engine, not a database: there are no joins, no cross-document transactions and no SQL, so it sits beside a primary datastore and keeping the two in step is work that belongs to you.

Zilliz

  • Pricing structure not publicly disclosed, requires sales contact
  • Operational complexity for self-hosted Milvus deployments
  • Learning curve for those unfamiliar with vector databases
  • Limited built-in analytics compared to some alternatives

Pricing, plan by plan

turbopuffer

$16/month
  • Launch$16/month
    • All database features
    • Multi-tenancy deployment
    • SOC2 & GDPR-ready DPA
  • Scale$256/month
    • Everything in Launch
    • HIPAA-ready BAA
    • Single Sign-On (SSO)
  • Enterprise$4096/month
    • Everything in Scale
    • Single-tenancy & BYOC deployment options
    • Private networking

Zilliz

Free

No published plan breakdown. See the Zilliz review.

Which should you pick?

Choose turbopuffer if

  • You need object storage architecture.
  • You also want namespaces.
turbopuffer review

Choose Zilliz if

  • You need vector indexing.
  • You want to start without paying.
  • You work on Cloud, Self-hosted.
  • You also want distributed architecture.
Zilliz review

Questions people ask

Is turbopuffer or Zilliz better?
Neither clearly leads. turbopuffer starts at $16/month and Zilliz at Free, and user ratings are close enough to be indistinguishable. Choose on capability and platform support.
Which is cheaper, turbopuffer or Zilliz?
Zilliz has a free tier; the other does not. Paid plans start at $16/month for turbopuffer and Free for Zilliz.
Does turbopuffer or Zilliz run on more platforms?
turbopuffer runs on Web. Zilliz runs on Cloud, Self-hosted.
Can I use Zilliz for free?
Yes. Zilliz has a free tier, so you can try it without paying. turbopuffer starts at $16/month.
What is turbopuffer best used for?
turbopuffer is most often used for a product with one search index per customer and thousands of customers, most of whose data is idle on any given day, very large corpora where holding every vector in memory is the dominant cost and occasional cold-query latency is acceptable, hybrid retrieval combining bm25 and vector search where running and synchronising two separate systems is the problem being solved, retrieval for agent and assistant products where indexes are created and destroyed frequently and per-index overhead must be near zero. Of those, a product with one search index per customer and thousands of customers, most of whose data is idle on any given day and very large corpora where holding every vector in memory is the dominant cost and occasional cold-query latency is acceptable are not what Zilliz is typically brought in for.
What can turbopuffer do that Zilliz cannot?
turbopuffer covers Object storage architecture, Namespaces, Vector search, Full-text search. Zilliz covers Vector indexing, Distributed architecture, SQL interface, Tensor support.

Answered from the vendors’ own pages

turbopuffer: Can I self-host turbopuffer?

There is no open source or community edition. Single-tenant and bring-your-own-cloud deployments exist as commercial arrangements, but there is no way to run it independently of the vendor.

Zilliz: What is the difference between Milvus and Zilliz Cloud?

Milvus is the open-source vector database that you can self-host. Zilliz Cloud is the fully managed service built on Milvus that removes operational overhead and handles scaling automatically.

Source
turbopuffer: How fast is it really?

Warm queries perform comparably to in-memory search engines. Cold queries, where data is not cached, have a documented p90 around 1,214 ms on a million documents. Write p90 is around 248 ms for a 512 KB upsert because writes go straight to object storage.

Zilliz: How many vectors can Zilliz handle?

Milvus and Zilliz Cloud can store and search billions of vectors through their distributed architecture that separates storage and compute layers.

Source
turbopuffer: Is it consistent?

Eventually consistent by default, with the vendor reporting that over 99.8% of queries return consistent data. Strong consistency can be requested per query at a latency cost. Large write bursts have a longer visibility delay while indexing catches up.

Zilliz: Is Milvus open-source?

Yes, Milvus is open-source under the Apache License 2.0 and is part of the LF AI & Data Foundation.

Source
turbopuffer: What is it best at?

Large numbers of namespaces where most are idle. The architecture makes cold data cheap to keep, which is exactly the shape of a multi-tenant product with a long tail of inactive customers.

Zilliz: What pricing does Zilliz Cloud offer?

Zilliz Cloud pricing is not publicly listed and requires contacting their team to discuss your specific scale and use case requirements.

Source
turbopuffer: What are the hard limits?

Up to 128 billion documents and 256 TB per namespace, 500 million documents per shard, 64 MiB per document, 10,752 dense vector dimensions, roughly 10,000 writes per second per namespace and a maximum result set of 10,000.

Share

Related pages

More on turbopuffer

More on Zilliz

Keep looking

Other head to heads