Vector & data
Embedding stores and search engines that back your RAG stack.
A vector database stores embeddings and finds the nearest matches at query time — the retrieval half of any RAG system. Qdrant, Weaviate, Milvus and Chroma are available as dedicated instances, and pgvector is the option that avoids running a second database at all.
Retrieval quality decides whether a knowledge assistant is useful or embarrassing, and retrieval quality is mostly a question of what happens in this layer: how documents are chunked, which embedding model is used, and whether the query can filter as well as compare.
Most projects start with the store embedded inside their application and move to a dedicated one for a specific reason — the corpus outgrew memory, several applications need the same index, or someone wants to query it directly. All of those are good reasons; starting with a separate instance on day one usually is not.
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
Vector search inside the database you already trust.
from $5.99/moDetails →
Meilisearch
Instant keyword search with optional semantic ranking.
from $5.99/moDetails →Choosing between these
Straight answers
Which one should I choose?
How much memory does a collection need?
Put your AI stack on your own box
Pick an app, pick a size, and have it running today. Month to month, cancel whenever.