Skip to Content
Relevan Beta is here 🎉
ResourcesCompareRelevan vs turbopuffer

Relevan vs turbopuffer

turbopuffer is a vector and full-text search database on object storage. Your team writes the search logic and tunes the ranking. Relevan is a managed search engine that tunes the retrieval of each index for you.

The facts about turbopuffer come from its public pages, as of October 2026. The sources are at the end of this page.

Summary

turbopufferRelevan
ApproachSearch infrastructureRelevance tuning for each index
RetrievalVector search, BM25 full-text search, filters, and regexKeyword retrieval on OpenSearch, with optional hybrid retrieval
RankingYour code, with rank fusion and optional managed rerankingConfiguration for each index, and an optional reranking model for each index
Who tunes relevanceYour teamRelevan
PricingUsage-based: storage, writes, and queries, with a monthly minimumMonthly plans, from a free plan

What turbopuffer does

  • turbopuffer stores all data in object storage, with a memory and SSD cache for queries.
  • It supports vector search (dense and sparse), BM25 full-text search, filters, regex search, and copy-on-write branches of a namespace.
  • The documentation says that turbopuffer is focused on first-stage retrieval. Your code fuses the results and reranks them. turbopuffer also offers rank fusion with RRF and native reranking with managed models.
  • To improve quality, the documentation tells you to build a test set of queries, measure nDCG, rewrite queries, and try chunking strategies.
  • It has multi-tenant, single-tenant, and bring-your-own-cloud deployments.

Vendor-reported figures: more than 1 trillion documents, 10 million writes per second, and 100,000 queries per second in production. For BM25 on 1 million documents, the documentation gives a p50 of 13 ms warm and 316 ms cold. turbopuffer says that it costs 10 times less than other options.

turbopuffer prices by use: up to $0.33 per GB of storage, up to $2.00 per GB written, and up to $1.00 per PB queried. The launch plan has a minimum of $16 a month. The enterprise plan starts at $4,096 a month.

What Relevan does

Relevan is a managed search engine for your own records, such as products, profiles, and documents. You send JSON records and a mapping. Relevan runs the OpenSearch index, and agents search it through an HTTP API.

  • Each search response tells the agent why each result matched, and suggests what to try next.
  • Relevan tunes retrieval and ranking for each index. Compare Relevan with other search tools explains what Relevan tunes, and when.
  • The feedback endpoint records clicks, LLM judgments, and conversions for each search.
  • Relevan prices by plan, from a free plan. See pricing.

The main difference

turbopuffer gives you fast and low-cost retrieval at a very large scale, and your team owns the relevance work. Relevan owns the relevance work for each index. This work covers field weights, BM25 parameters, the threshold for long queries, and hybrid retrieval. It also covers a reranking model for each index.

With turbopuffer, your team decides when and how to change ranking. With Relevan, Relevan assesses an index from its documents and its search logs, and applies the changes that the evidence supports.

When to choose turbopuffer

  • You have a search or ML team that wants full control of ranking in its own code.
  • You store very large data sets, or many tenants with one namespace each.
  • Low latency is critical. In an October 2026 test from one machine, Relevan had a p50 of 144 ms on a product catalog. turbopuffer reports 13 ms warm for BM25.
  • You need bring-your-own-cloud deployment.

When to choose Relevan

  • You do not have a team to build and tune ranking.
  • You want each index to start with a tuned configuration, and to change from the evidence in its own search logs.
  • Your agents need an explanation of each match and a suggestion for the next search.

Sources

Read on October 29, 2026:

Last updated on
Relevan vs OpenSearchRelevan vs ZeroEntropy