turbopuffer
@turbopuffer
{vector, full-text} search engine built on object storage. fast, cheap, 1T scale. powers Anthropic, Cursor, Notion, and more
5 Following 14.5K Followers
tpuf BM25 index tracks matching doc IDs per term:
[1, 8, 32, ... 1048576]
instead of IDs (big), we store the deltas (small):
[1, 7, 24, ... 16]
we prefix sum the deltas at query time to rebuild IDs
applying @lemire's SIMD speedup gave us up to 13% faster text search on ARM
Show more
new: highlighting
extract the text fragments most relevant to a query
→ highlight matches in your search results UI
→ minimize context passed to your LLM
docs (and playground):
Show more
Nemotron 3 Embed is now available on tpuf native embeddings via our partner @baseten
contact us for beta access, full model list here:
Today we released Nemotron 3 Embed 8B and it reached #1# overall on RTEB 🏆
RTEB benchmarks retrieval accuracy across real-world tasks. Better retrieval gives agents more relevant context, helping improve response accuracy.
Show more
now in beta: native embeddings in tpuf
embedding is the most painful part of puffing. we want to make it easy
you can now convert chunks to vectors as you read and write to turbopuffer, without extra calls to an embedding model provider API
docs:
Show more
TP SUMMER 26: the beach puff 🌞
now live on
Legora searches 2B+ legal documents on turbopuffer
→ strict per-matter data isolation with CMEK
→ 10x lower tail latency than Postgres
→ 98% avg recall@10
new: i8 vectors
f32: 4 bytes/dim
i8: 1 byte/dim
4x fewer bytes → 75% lower storage and query costs + faster queries
when embedded with a quantization-aware model (e.g. voyage-4-large) trained on i8 vectors, recall loss can be ~0!
docs:
Show more
Atlassian's cross-product AI platform, Rovo, searches 5B+ documents on turbopuffer BYOC
→ 19% increase in search quality
→ 60ms p90 latency
→ 96% average recall@10
a year ago, ~98% of tpuf queries were vector ANN
last 30d:
64% vector ANN
19% full-text BM25
13% filter-only
3% aggregate
1% other (sparse vector, exact kNN, ...)
new: rerank_by
before, you'd implement rank fusion client-side. now, a little QoL upgrade, especially nice for large result sets
docs:
new: branching
create an instant, copy-on-write clone of a tpuf namespace
→ constant-time (440ms p50, ~1s p99)
→ fully independent
→ unlimited branches, unlimited branch depth
docs:
Show more
tpuf quantizes vectors to improve perf (RaBitQ)
the algo randomly rotates vectors, and we were using matmul at O(d²) space & time, brutal at high dims. 10k = 400MB in RAM!
we rebuilt the rotation using FWHT at O(d) space & O(d log d) time. ~no recall loss, 10k = only 5kB in RAM
Show more
new in turbopufer: the Fuzzy filter
typo-tolerant substring matching with a configurable edit distance, so you can puff (or puf) even when you spell it wrong
docs:
Show more
puff
turbopuffer crossed $100M run-rate in March. 19mo after $1M. Profitable & <$1M raised.
Cursor・Anthropic・Notion・Cognition・Harvey・Bridgewater・Ramp・Linear・Legora・Superhuman・Atlassian・Granola
We’d be nowhere without them. We work like hell to exceed their expectations.
Show more
