turbopuffer
x
TLDR
turbopuffer x TLDR
search every byte with turbopuffer
vector and full-text search built on object storage
fast, 10x cheaper, and extremely scalable
About turbopuffer
turbopuffer is the search engine to connect large amounts of unstructured data to AI. Built from first principles on object storage with an intelligent cache layer, turbopuffer is just as fast as in-memory search engines but orders of magnitude cheaper to run.
We index 1T+ documents for leading AI companies:
Cost calculator
| Expected cost | |||
|---|---|---|---|
| Storage | ≤ $0.33 per GB/GB | 1M docs (~3GB) | |
| Writes | ≤ $2.00 per GB/GB | 1M writes, ~24 WPM (~5GB) | |
| Queries | ≤ $1.00 per PB/PB | 1M queries, ~24 QPM (~1PB) | |
| Namespaces | 100K docs per namespace/NS | 10 | Included |
| Storage | |
| Writes | |
| Queries | |
| Namespaces | Included |
Estimated cost
(min $16/mo for launch plan)
/month
Workload1024 dimensions, 10M docs, ~40GB
p50
14ms
874ms
p90
17ms
1214ms
p99
27ms
1686ms
Warm namespace
Cold namespace
Approach (8 QPS with topk=10)
turbopuffer/tpuf-benchmark
