The deciding difference is operational: pgvector can run on infrastructure you control, turbopuffer cannot. If that constraint is real for you, it settles the choice before anything else is considered.
Vector search inside the Postgres you already run.
Correct default under roughly ten million vectors. Reach for a dedicated store when it stops being.
Vector and full-text search on object storage — very cheap at rest.
| pgvector | turbopuffer | |
|---|---|---|
| Licence | Open source | Proprietary |
| Pricing | open source | usage based |
| Self-hostable | Yes | No |
| Languages | Any language | Python, TypeScript |
| Install | — | — |
| Keys required | None | TURBOPUFFER_API_KEY |
The deciding difference is operational: pgvector can run on infrastructure you control, turbopuffer cannot. If that constraint is real for you, it settles the choice before anything else is considered. pgvector: Vector search inside the Postgres you already run. turbopuffer: Vector and full-text search on object storage — very cheap at rest.
pgvector can run on your own infrastructure. turbopuffer is a hosted service only.
pgvector is open source (open source). turbopuffer is proprietary (usage based).
Every page here answers to Accept: text/markdown and returns the same content at roughly a tenth the tokens. No separate site, no toggle — same URL.
curl -s -H "Accept: text/markdown" https://newagent.build/compare/pgvector-vs-turbopuffer