Pyyan / Compare / turbopuffer vs pgvector vs Qdrant vs Pinecone
turbopuffer vs pgvector vs Qdrant vs Pinecone
4 of 5Vector Databases · verified 13 Aug 2026
| Specification | turbopuffer | pgvector | Qdrant | Pinecone |
|---|---|---|---|---|
| Summary | Vectors on object storage, priced accordingly. | Vectors inside the database you already run. | The fastest of the purpose-built stores. | Managed, and the smoothest to operate. |
| Index | Proprietary, object storage | HNSW, IVFFlat | HNSW | Proprietary |
| Hybrid search | Yes | Yes, with SQL | Yes | Yes |
| Hosting | Managed only | Self-host or any managed Postgres | Self-host and managed | Managed only |
| Licence | Proprietary | PostgreSQL | Apache 2.0 | Proprietary |
| p50 latency | ~20ms | ~15ms | 4ms | <10ms |
| Category | Vector Databases | Vector Databases | Vector Databases | Vector Databases |
| Official | turbopuffer ↗ | PostgreSQL ↗ | Qdrant ↗ | Pinecone ↗ |
Highlighted rows are where these differ.
turbopuffer
- Built on object storage rather than RAM
- Notion cut search costs ~60% moving from Pinecone Serverless
Best for large corpora where cost dominates.
pgvector
- One database for vectors, rows and joins
- The honest default: reach for something else only when this stops working
Best for almost everyone, until scale says otherwise.
Qdrant
- Written in Rust; 10 to 25% faster than Weaviate or Milvus on common workloads
- Strong filtering alongside vector search
- Self-host or managed, same engine
Best for latency-sensitive retrieval.
Pinecone
- Sub-10ms p50 with nothing to maintain
- Costs draw scrutiny at scale; Notion moved away and cut spend ~60%
Best for teams who do not want to run a database.