Postgres + pgvector hosting
Vector search inside the database you already trust.
Postgres + pgvector hosting at AIHostBox means a dedicated, single-tenant Postgres + pgvector instance on our infrastructure: we install it, pin it to a reviewed release, put it behind TLS on your domain, back it up nightly and keep it patched. Plans start at $5.99 per month and the admin account, the data and any API keys are yours.
pgvector is a PostgreSQL extension, created and maintained by Andrew Kane, that adds a vector type and similarity search to the database you may already run. It stores vectors of up to 16,000 dimensions, builds HNSW and IVFFlat indexes, and provides L2, inner product, cosine, L1, Hamming and Jaccard distance operators. Half-precision, binary and sparse vector types cut storage for large embeddings, and since version 0.8 iterative index scans stop filtered queries from returning too few rows.
The case for it is that it is still Postgres: transactions, foreign keys, row-level security, full-text search and ordinary pg_dump backups all apply to your embeddings. That is also why it belongs on a server. Your application, a RAG framework or several services connect to it over the network at all hours, and HNSW indexes are searched fastest when they sit in memory, so the database wants RAM of its own rather than a laptop shared with a browser.
Postgres + pgvector hosting plans
Every plan is a dedicated single-tenant instance with NVMe storage, nightly off-box backups, TLS and managed patching. Quarterly, semi-annual and annual terms are discounted at checkout. Model tokens are not included — you bring your own key.
How we run Postgres + pgvector
A single PostgreSQL server from the official pgvector/pgvector image, with the vector extension enabled, data on NVMe and remote connections on port 5432 over TLS only.
What we configure for you
- CREATE EXTENSION vector in your database, with your application role and password
- shared_buffers and effective_cache_size sized to the plan’s RAM
- maintenance_work_mem raised so HNSW index builds fit in memory
- TLS required for remote connections, with SCRAM password authentication in pg_hba.conf
- Nightly off-box backup of the database, and a pg_dump on request for your own copy
Which plan fits
Starter (2 GB, 1 vCPU) suits a prototype or up to roughly 150,000 768-dimension vectors. Plus (4 GB, 2 vCPU) handles a few hundred thousand 1,536-dimension vectors with the HNSW index in memory. Pro (8 GB, 3 vCPU) is for around a million 768-dimension vectors or heavy concurrent querying; storing them as halfvec roughly halves the memory needed.
The honest limits
Approximate indexes trade recall for speed, so ef_search and list counts need tuning against your data; well beyond a few million vectors on one node, a dedicated engine such as Qdrant or Milvus usually scales more comfortably.
What people run Postgres + pgvector for
Avoid a second store
Keep vectors beside your relational data.
Filtered similarity
A WHERE clause and a similarity ordering in the same query.
RAG store for a framework
Dify, Flowise, LangChain and LlamaIndex all ship a pgvector store you can point at this database.
Hybrid search
Combine Postgres full-text ranking with vector similarity for better retrieval.
Licence and version policy
| Item | Position |
|---|---|
| Upstream licence | PostgreSQL License |
| Sold as managed hosting | Yes |
| Tenancy | Single tenant — one customer per instance |
| Patching | Critical advisories within 24 hours, high severity within 72 |
Our full position on licences, including the applications we decline to sell, is on the licensing page.
Reading on Postgres + pgvector
Works well with Postgres + pgvector
Dify
A visual studio for building LLM apps, agents and RAG pipelines.
from $18.99/moDetails →
Flowise
Drag-and-drop LLM flows, agents and chatbots you can embed anywhere.
from $11.99/moDetails →
Ollama
Your private LLM server — pull a model, get an API, keep every token on your own box.
from $16.99/moDetails →More Vector & data we host
Qdrant
A fast vector database with real filtering, written in Rust.
from $5.99/moDetails →
Weaviate
A vector database with hybrid search and built-in vectorization modules.
from $11.99/moDetails →
Milvus
Vector search built for very large collections.
from $18.99/moDetails →
Chroma
A lightweight embedding store that gets out of the way.
from $5.99/moDetails →Postgres + pgvector hosting: your questions
Which pgvector and Postgres versions do you run?
How many dimensions can I index?
HNSW or IVFFlat: which index should I use?
Can I connect with psql and my usual drivers?
How do I import existing embeddings?
Run Postgres + pgvector without running a server
A dedicated Postgres + pgvector instance, deployed and maintained by us. Month to month, cancel whenever.