turbopuffer

turbopuffer belongs in the catalog because it represents a distinctive infrastructure bet inside the retrieval market. The official positioning is not just generic vector search. It is serverless vector and full-text search built on object storage, with the explicit promise of being far cheaper than alternatives. That matters because many teams building search and RAG systems are no longer just comparing API ergonomics. They are comparing the structural cost model of the underlying system and asking whether a newer architecture can materially reduce spend without giving up production utility.

It also deserves inclusion because the pricing page is unusually explicit about billing mechanics. turbopuffer publicly shows a Launch plan with a $64 per month minimum usage, explains that actual usage beyond the minimum is billed separately, and documents details such as a 10 KB minimum write charge, batch write discounts up to 50 percent, a 1.28 GB minimum queried size, and steep marginal discounts for larger queried datasets. That kind of pricing transparency is rare in search infrastructure and makes turbopuffer especially worth cataloging for serious technical buyers.