turbopuffer review

Q: How do I install turbopuffer?

Run pip install turbopuffer. The install section above has the other options from the project's README.

turbopuffer is a hosted vector and full-text search database built on object storage, with an official Python client library.

Updated Nov 1, 2026. Review written Oct 22, 2026 from the project's own sources; usage data refreshed every Monday.

TL;DR

turbopuffer is a vector and full-text search database built on object storage, reached through an HTTP API. The official Python library gives Python 3.10+ applications typed sync and async clients for it. It fits developers building semantic search, retrieval, or recommendation features who want search backed by object storage with a memory/SSD cache.

183GitHub stars
+5new stars, week of Oct 25
~3.4Mdownloads a week, +32% vs 4-week avg
#18in RAG, data and memory this week

turbopuffer at a glance

Category: RAG, data and memory
Maintained by: turbopuffer (organization, @turbopuffer on GitHub)
License: MIT
Language: Python, Shell, Dockerfile
Repository: github.com/turbopuffer/turbopuffer-python
Website: turbopuffer.com
npm package: @turbopuffer/turbopuffer 2.11.1
PyPI package: turbopuffer 2.14.0
Latest release: v2.14.0 on GitHub, Oct 21, 2026
First published: Dec 12, 2023

What turbopuffer does

turbopuffer runs approximate nearest neighbor vector queries with attribute filters and a top_k limit.
turbopuffer supports BM25 full-text search on attributes, alongside hybrid search and metadata filtering.
turbopuffer stores documents in namespaces that you write to by upserting rows with vectors and a distance metric such as cosine distance.
turbopuffer's Python client offers synchronous and asynchronous clients with type definitions for all request params and response fields.
turbopuffer's client auto-paginates list methods, retries certain errors with exponential backoff, and times out requests after 1 minute by default.
turbopuffer's Python client supports reranking with a model as of version 2.13.0.
turbopuffer lets you drop attributes from a namespace schema and create a namespace with a write that contains no documents.

How it works

turbopuffer is a hosted service. Your client calls its HTTP API, which serves queries from a memory/SSD cache in front of object storage (S3). The Python library wraps that API using httpx, or optionally aiohttp for async. You create a client with a region and an API key, open a namespace, then write rows and run vector or full-text queries.

How to install and start using turbopuffer

With pip

pip install turbopuffer

With pip and the aiohttp backend

pip install turbopuffer[aiohttp]

Requirements

Python 3.10 or later
A turbopuffer API key, read from TURBOPUFFER_API_KEY by default
A turbopuffer region chosen when creating the client

First steps

Sign up for turbopuffer and get an API key.
Add TURBOPUFFER_API_KEY to your .env file with python-dotenv so the key stays out of source control.
Create a Turbopuffer client with your region, for example gcp-us-central1, and open a namespace with tpuf.namespace().
Run ns.query() with rank_by set to a vector ANN query or a BM25 text query.

Who turbopuffer is for

Python developers adding vector or full-text search to AI applications
Teams building semantic search or recommendation systems on a hosted search database
Async Python services that need a typed search client

Limitations to know

turbopuffer is a hosted API, so the client needs network access and an account API key to do anything.
turbopuffer's Python library requires Python 3.10 or later.
turbopuffer's Python client does not use compression by default, which favors CPU efficiency over bandwidth.

Is turbopuffer growing? Usage data

In the week of Oct 25–31, 2026, turbopuffer ranked #170 on the Devtoolwatch Board and #18 in RAG, data and memory, with a momentum score of 49 out of 100. Its main package had about 3.4M downloads that week, +32% against its own 4-week average.

New GitHub stars per week for the last 12 weeks, from GitHub's stargazer history (today's stargazers by the week they starred).

Show the weekly numbers

New GitHub stars per week for turbopuffer
Week of	New stars
Aug 9, 2026	0
Aug 16, 2026	2
Aug 23, 2026	1
Aug 30, 2026	4
Sep 6, 2026	1
Sep 13, 2026	3
Sep 20, 2026	0
Sep 27, 2026	3
Oct 4, 2026	2
Oct 11, 2026	4
Oct 18, 2026	2
Oct 25, 2026	5

On Hacker News

RIP, vector database, 405 points and 116 comments (Oct 1, 2026)
Turbopuffer: Fast search on object storage, 379 points and 64 comments (Jul 9, 2024)
A distributed queue in a single JSON file on object storage, 171 points and 55 comments (Feb 21, 2026)
ANN v3: 200ms p99 query latency over 100B vectors, 109 points and 47 comments (Jan 21, 2026)
How to ship a database every day, 71 points and 10 comments (Aug 17, 2026)

The Board ranks tools by four signals each week: new stars, new stars for the repo's size, added downloads and download growth beyond the market. How the ranking works.

turbopuffer alternatives

Other RAG, data and memory tools on the Board in the week of Oct 25–31, 2026, by momentum:

RAG, data and memory ranked by momentum, the week of Oct 25–31, 2026
Rank	Tool	Stars	Downloads/wk
#1	XERJXERJ is a local-first search engine and MCP server for AI agents, packaged as one Rust binary with an Elasticsearch-compatible API.	17.9k	–
#2	MemvidMemvid is a single-file memory layer for AI agents that packages data, embeddings, search indexes, and metadata into one portable .mv2 file.	17.3k	–
#3	LEANNLEANN is a Python vector database and RAG library that recomputes embeddings on demand to keep indexes small on personal devices.	13.7k	–
#4	GraphifyGraphify is an AI coding assistant skill that turns a codebase, with its docs, PDFs, images and videos, into a queryable knowledge graph.	140k	480.7k
#5	MarkItDownMarkItDown is a lightweight Python utility from Microsoft that converts PDFs, Office documents, images, audio and other files to…	193.2k	4.7M
#6	FirecrawlFirecrawl is an API and open source project to search, scrape, crawl, and interact with web pages, returning LLM-ready markdown or JSON…	197.4k	2.9M

All RAG, data and memory tools, ranked

Recent turbopuffer releases

v2.14.0, Oct 21, 2026
v2.13.0, Oct 13, 2026
v2.12.0, Oct 9, 2026
v2.11.0, Oct 7, 2026
v2.10.2, Sep 23, 2026

turbopuffer FAQ

Is turbopuffer free and open source?

Yes. turbopuffer's source code is on GitHub (turbopuffer/turbopuffer-python) under the MIT license, so you can use it for free. Its website also links to a pricing page (https://turbopuffer.com/pricing) for any paid plans.

How do I install turbopuffer?

Run pip install turbopuffer. The install section above has the other options from the project's README.

Is turbopuffer still maintained?

Yes. The turbopuffer/turbopuffer-python repository was last updated on Oct 21, 2026, and the latest release, v2.14.0 on GitHub, came out on Oct 21, 2026.

How popular is turbopuffer?

turbopuffer has 183 GitHub stars as of Nov 1, 2026. It gained 5 new stars in the week of Oct 25–31, 2026, against a 12-week average of 2 a week. Its main package had about 3.4M downloads that week, +32% against its own 4-week average. That made it #18 among RAG, data and memory tools on the Devtoolwatch Board.

turbopuffer vs pgvector: how do they compare?

turbopuffer and pgvector are both tracked by Devtoolwatch as RAG, data and memory tools. turbopuffer has 183 GitHub stars and pgvector has 23.5k. In the week of Oct 25–31, 2026, turbopuffer's main package had 3.4M downloads and pgvector's had 9.4M. pgvector is an open-source Postgres extension for exact and approximate vector similarity search, storing vectors alongside the rest of your data. pgvector review.

turbopuffer vs Qdrant: how do they compare?

turbopuffer and Qdrant are both tracked by Devtoolwatch as RAG, data and memory tools. turbopuffer has 183 GitHub stars and Qdrant has 35.4k. In the week of Oct 25–31, 2026, turbopuffer's main package had 3.4M downloads and Qdrant's had 4.5M. Qdrant is an open-source vector similarity search engine and vector database written in Rust, with filtering on JSON payloads. Qdrant review.

What are the best alternatives to turbopuffer?

Other RAG, data and memory tools on the Devtoolwatch Board in the week of Oct 25–31, 2026, by momentum: XERJ (#1), Memvid (#2), LEANN (#3), Graphify (#4), MarkItDown (#5). The alternatives table above has their numbers.

Is turbopuffer a vector database?

Yes. turbopuffer is a vector and full-text search database built on object storage. It supports vector search, full-text search, hybrid search, and metadata filtering.

What is turbopuffer used for?

turbopuffer is used for search in AI applications, semantic search, recommendation systems, and other similarity search use cases. Queries can rank by vector ANN or by BM25 text relevance.

How does turbopuffer work?

turbopuffer stores data in object storage (S3) with a memory/SSD cache in front of it. Clients reach it through an HTTP API, organizing data into namespaces they write to and query.

Does the turbopuffer Python client support async?

Yes. turbopuffer's library provides AsyncTurbopuffer, which you use with await. Functionality matches the sync client. You can also install the aiohttp extra and pass DefaultAioHttpClient() for the HTTP backend.

Is there a TypeScript client for turbopuffer?

Yes. turbopuffer publishes an official TypeScript library on npm as @turbopuffer/turbopuffer.

Sources

turbopuffer README badge

Maintainers can show the tool's weekly Board rank in their README:

[![Devtoolwatch](https://index.devtoolwatch.com/badge/turbopuffer.svg)](https://index.devtoolwatch.com/tools/turbopuffer/)

How this page is made: Devtoolwatch writes each review from the project's README, releases, package metadata and website, with an AI model, then checks every claim against the source it cites and drops what the sources don't support. The numbers come from Devtoolwatch's daily GitHub snapshots, GitHub's stargazer history, and the npm and PyPI download APIs. Spotted an error? Email [email protected].