banner
LightNews
turbopuffer.bsky.social
turbopuffer
@turbopuffer.bsky.social
150 followers 1 following 38 posts
search every byte 🔍 {vector, full-text} search engine built from first principles on object storage. 10x cheaper, scales to 100B. powers Notion, Cursor, Linear
Posts Media Videos Starter Packs
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · 25d
august changelog

turbopuffer.com/docs/roadmap
2
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Sep 3
"Having our data in turbopuffer makes us feel ready for whatever comes next in AI-enabled software. It gives us the confidence to adapt quickly."
— Tristan Homsi, co-founder of Readwise

turbopuffer.com/customers/r...
Readwise → turbopuffer, the use-case that inspired turbopuffer's creation
serverless vector and full-text search built from first principles on object storage: fast, 10x cheaper, and extremely scalable
turbopuffer.com
1 1
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Aug 26
new: native facet support with group_by
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Aug 7
july changelog

turbopuffer.com/changelog
1
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Aug 6
cold full-text search p90 decreased ~30%
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Aug 6
full-text search p90 dropped ~50%
1
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Aug 6
every few months we re-run the landing page benchmarks against prod

vector search p90 down ~50%
1
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Aug 5
ripgrep on object storage?
2
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jul 23
our ruby client just dropped 💎
2
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jul 22
by popular demand, now introducing one-dimensional vectors 🎈
1 4
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jul 17
"turbopuffer makes it too easy to build state-of-the-art AI apps" – Notion

turbopuffer.com/customers/n...
Notion
serverless vector and full-text search built from first principles on object storage: fast, 10x cheaper, and extremely scalable
turbopuffer.com
1
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jul 11
new GCP regions. now puffin' in Toronto and Singapore

turbopuffer.com/docs/regions
2
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jul 9
june changelog

turbopuffer.com/docs/roadmap
2
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jul 7
see pricing page for more turbopuffer.com/pricing
Pricing
serverless vector and full-text search built from first principles on object storage: fast, 10x cheaper, and extremely scalable
turbopuffer.com
1
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jul 7
we reduced query pricing for large namespaces by up to 80%

puff more, pay less
1 2
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jul 3
docs here: turbopuffer.com/docs/write#c...
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jul 3
S3’s conditional writes were the key ingredient that made our architecture possible

now it's our time to pay it forward
1 1
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jun 26
"Their responsiveness and shipping velocity make us feel like we are their only customer." – Tom Moor, Head of Engineering at Linear

turbopuffer.com/customers/l...
2
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Jun 6
changelog for May 🐡🎉

turbopuffer.com/docs/roadmap
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · May 14
turbopuffer is generally available. petabytes in prod.

turbopuffer.com/
4 12
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · May 12
🇨🇦 🏔️ 🗽

turbopuffer.com/about
1
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · May 7
changelog for April ✨🐡

turbopuffer.com/docs/roadmap
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · May 6
important considerations include:
- sync vs async I/O
- buffered vs direct I/O
- prioritization and scheduling
- request depth, size, and pattern (sequential vs random)
- striping and alignment

to learn more about the topic, we recommend starting with vldb.org/pvldb/vol16/...
vldb.org
1
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · May 6
modern NVMe drives are blazing fast, but software needs to be architected from the ground up to take full advantage of them

we recently rolled out a new I/O scheduler that uses Direct I/O to maximize performance of our SSD caches, avoiding page cache cpu and locking overhead
2 2 3
turbopuffer.bsky.social
turbopuffer @turbopuffer.bsky.social · Apr 28
"After rolling out Turbopuffer, the searchDocs step in our retrieval pipeline went from being the slowest to the fastest."

🚀 🐡 Read how turbopuffer improved search response time for ref.tools: ref.tools/use-case/tur...
Ref.
ref.tools
1