turbopuffer
Vector and full-text search built on object storage, priced far below memory-based stores.
why this verdict
Keep it — Anthropic has not replaced it
Anthropic's version overlaps, but does not finish the dev tools job, so this one is still worth keeping open.
- A named launch, not a vibe named, unlinked
- Pgvector and bundled platform search covering retrieval without a separate vendor. Anthropic, April 16, 2025 — no announcement link recorded yet.
- How much of the job it covers not the job editorial call
- Parts of it. The job still needs the tool to get finished.
- Is there a free way to do it? yes
- 2 of 3 listed replacements have a usable free tier: pgvector and Qdrant.
- What the call is worth nothing to cancel
- No paid entry tier tracked, so there is no subscription to cancel.
- Threatened by
- Anthropic
- Since
- April 16, 2025
- List price
- free
- Per year
- —
The backstory
turbopuffer's design decision is to keep vectors on object storage rather than in RAM, which trades a little latency for an order of magnitude less cost, and caches hot namespaces to claw most of the latency back. That is the right trade for the shape retrieval actually takes, where a small fraction of data is queried constantly and the rest almost never. It won adoption at companies with very large corpora precisely where memory-resident vector databases become financially absurd.
Escape hatches
Free and already in your database
github.com open_in_newFaster for smaller, latency-critical workloads
qdrant.tech open_in_newFully managed with a longer track record
pinecone.io open_in_new2 of 3 replacements have a usable free tier.
