datastores.ai

Best Vector Databases for OpenAI Embeddings

OpenAI's embedding models (text-embedding-3-small, text-embedding-3-large, and the legacy ada-002) are the most widely used for generating vector representations of text. Choosing the right vector database to store and query these embeddings is essential for building performant AI applications. The databases below all work seamlessly with OpenAI embeddings and support the common 1536 or 3072 dimensions used by these models.

15 databases compatible with OpenAI

Pinecone

Pinecone

Serverless vector database for AI at scale

CloudServerless
Qdrant

Qdrant

High-performance vector search engine in Rust

Open SourceRust
ChromaDB

ChromaDB

The AI-native open-source embedding database

Open SourceEmbedded
pgvector

pgvector

Vector search for PostgreSQL

Open SourcePostgreSQL
Supabase Vector

Supabase Vector

pgvector on Supabase — vectors in your Postgres

Open SourcePostgreSQLBaaS
Azure AI Search

Azure AI Search

Microsoft's enterprise vector + full-text search service

CloudAzure
Deep Lake

Deep Lake

Multi-modal AI data lake with vector search

HybridMulti-Modal
Weaviate

Weaviate

AI-native vector database with built-in vectorizers

Open SourceHybrid Search
Milvus

Milvus

Distributed vector database built for scale

Open SourceDistributed
Redis Vector

Redis Vector

Sub-millisecond vector search in memory

TraditionalIn-Memory
MongoDB Atlas Vector Search

MongoDB Atlas Vector Search

Vector search built into the #1 document database

TraditionalDocument DB
Zilliz Cloud

Zilliz Cloud

Managed Milvus with enterprise scalability

CloudEnterprise
Upstash Vector

Upstash Vector

Serverless vector database with per-request pricing

CloudServerlessEdge
Elasticsearch

Elasticsearch

Distributed search engine with vector capabilities

TraditionalSearch Engine
Google Vertex AI Vector Search

Google Vertex AI Vector Search

Google-scale vector search on GCP

CloudGCP

Why use OpenAI with a vector database?

OpenAI's embedding models (text-embedding-3-small, text-embedding-3-large, and the legacy ada-002) are the most widely used for generating vector representations of text. Choosing the right vector database to store and query these embeddings is essential for building performant AI applications. The databases below all work seamlessly with OpenAI embeddings and support the common 1536 or 3072 dimensions used by these models.

How to get started with OpenAI

  1. 1Generate embeddings using OpenAI's API: openai.embeddings.create(model='text-embedding-3-small', input=text)
  2. 2Choose a vector database and create a collection with the matching dimension (1536 or 3072)
  3. 3Store embeddings with their source text as metadata for retrieval
  4. 4Query with new embeddings to find semantically similar content for RAG or search

FAQ — OpenAI & Vector Databases

Which vector database is best for OpenAI embeddings?

Pinecone is the most popular pairing with OpenAI embeddings for production. ChromaDB is great for prototyping, and Qdrant offers excellent performance with a free tier. All support OpenAI's 1536 and 3072 dimensions.

What dimension should I use for OpenAI embeddings?

text-embedding-3-small produces 1536-dimension vectors, and text-embedding-3-large produces 3072 dimensions. All vector databases on this page support both sizes.

Can I use OpenAI embeddings with a free vector database?

Yes. ChromaDB (local, open-source), pgvector (free PostgreSQL extension), and Qdrant (free cloud tier with 1GB) all work perfectly with OpenAI embeddings at no database cost.

Explore more

LangChainLlamaIndexPythonRAG PipelinesVector DB FinderBenchmarksCompare Databases