turbopuffer
Vector and full-text search database built on object storage — fast, scalable, and 10x cheaper than alternatives.
| What is it | Vector and full-text search database built on object storage — fast, scalable, and 10x cheaper than alternatives. |
|---|---|
| Pricing | Paid — from $16/mo |
| Free tier | No |
| Platform | API |
| API | Yes |
| Best for | semantic search for AI applications, recommendation systems |
| Domain registered | 2023 |
Data updated Aug. 1, 2026
What does turbopuffer do?
Turbopuffer is a search database that handles both vector and full-text search, but it doesn't run on the usual expensive infrastructure. Instead, it's built directly on top of object storage like S3. That means it can store and query billions of vectors without the high costs of traditional vector databases. The company claims it's about 10x cheaper than alternatives, and the numbers back it up — they're handling over 4 trillion documents and 10 million writes per second in production for companies like Cursor, Notion, and Anthropic.
How does it work? You upload your data (vectors, text, metadata) into a namespace, then query it using vector similarity, BM25 full-text search, or a hybrid of both. Turbopuffer automatically scales, uses a memory/SSD cache for fast queries, and stores everything durably on object storage. The latency is low — sub-10ms p50 for warm queries — and you can filter by metadata, pin namespaces for speed, and even branch namespaces for experimentation. The pricing starts at $16/month for a launch plan, which is surprisingly affordable for a production-grade search database.
Who is this for? Developers building AI applications that need semantic search, recommendation systems, or retrieval-augmented generation (RAG). It's also great for any app that needs fast, scalable full-text search without the complexity of managing a cluster. If you're tired of paying huge bills for vector databases or dealing with infrastructure headaches, turbopuffer is worth a serious look.
Key features
What makes it stand outWho is turbopuffer for?
Who benefits most from this toolPricing
launch
- Includes all database features
Trust & presence
Gallery
Click any image to enlargeAlternatives in Data Mining
API for document parsing, data extraction, AI Q&A, and academic paper search
AI document processing tool that extracts structured JSON data from PDFs and images in seconds
AI-powered Chrome extension that scrapes data from websites, PDFs, and images using natural language commands.
AI-powered web scraper — extract clean, structured data from any website at any scale.
AI platform that automates complex document workflows — extracting, analyzing, and validating data from emails, PDFs, and scans.
Enterprise AI platform for automated document processing, data extraction, and intelligent search.
AI-powered API that extracts structured data (like prices, dates, names) from any text using natural language descriptions.
API for real-time search results and web scraping — get structured JSON data for AI applications starting at $0.0008/request.
Similar tools
Serverless vector database for AWS – pay per request, scale from prototype to production with a few lines of code.
Cloud browser API for AI agents — instant headless browsers for scraping, testing, and automation without infrastructure.
AI-powered search service for applications — add semantic search, vector search, and document intelligence to your apps
High-performance vector database for AI workloads — fast similarity search, metadata filtering, and optional compression.
Vector + graph database platform for multimodal AI data, powering GenAI pipelines and intelligent agents
AI-native database for LLM applications — provides fast hybrid search across vectors, text, and tensors.