Vector Databases

Fresh

Mem0 supports 27+ vector database providers. Qdrant is the default.

Language Support

Python supports all providers. TypeScript currently supports Qdrant, Redis, Valkey, Vectorize, and in-memory vector database.

Supported Providers

Provider	Python	TypeScript	Type
Qdrant	Yes	Yes	Default
Redis	Yes	Yes	In-memory
Valkey	Yes	Yes	Redis fork
Vectorize	Yes	Yes	Managed
In-memory	Yes	Yes	Local
Chroma	Yes	No	Open source
PGVector	Yes	No	PostgreSQL
Milvus	Yes	No	Cloud-native
Pinecone	Yes	No	Managed
MongoDB	Yes	No	Atlas
Azure AI Search	Yes	No	Enterprise
Azure MySQL	Yes	No	Enterprise
Elasticsearch	Yes	No	Full-text + vector
OpenSearch	Yes	No	AWS managed
Supabase	Yes	No	PostgreSQL
Upstash Vector	Yes	No	Serverless
Vertex AI	Yes	No	Google Cloud
Weaviate	Yes	No	Graph + vector
FAISS	Yes	No	Facebook AI
LangChain	Yes	No	Framework bridge
Baidu VectorDB	Yes	No	Baidu Cloud
Cassandra	Yes	No	Apache
Amazon S3 Vectors	Yes	No	AWS
Databricks	Yes	No	Lakehouse
Neptune Analytics	Yes	No	AWS graph
Turbopuffer	Yes	No	Serverless

Configuration

from mem0 import Memory

config = {
    "vector_store": {
        "provider": "qdrant",
        "config": {
            "host": "localhost",
            "port": 6333,
            "collection_name": "memories",
        }
    }
}

memory = Memory.from_config(config)

Provider Examples

Qdrant (Default)

config = {
    "vector_store": {
        "provider": "qdrant",
        "config": {
            "host": "localhost",
            "port": 6333,
        }
    }
}

Chroma

config = {
    "vector_store": {
        "provider": "chroma",
        "config": {
            "collection_name": "memories",
            "path": "/path/to/chroma/db",
        }
    }
}

PGVector

config = {
    "vector_store": {
        "provider": "pgvector",
        "config": {
            "dbname": "mem0",
            "user": "postgres",
            "password": "your-password",
            "host": "localhost",
            "port": 5432,
        }
    }
}

Pinecone

config = {
    "vector_store": {
        "provider": "pinecone",
        "config": {
            "api_key": "your-pinecone-key",
            "environment": "us-east-1-aws",
            "index_name": "memories",
        }
    }
}

Supabase

config = {
    "vector_store": {
        "provider": "supabase",
        "config": {
            "url": "https://your-project.supabase.co",
            "key": "your-service-role-key",
            "table_name": "memories",
        }
    }
}

Redis

config = {
    "vector_store": {
        "provider": "redis",
        "config": {
            "redis_url": "redis://localhost:6379",
            "collection_name": "memories",
        }
    }
}

Milvus

config = {
    "vector_store": {
        "provider": "milvus",
        "config": {
            "url": "http://localhost:19530",
            "collection_name": "memories",
        }
    }
}

Elasticsearch

config = {
    "vector_store": {
        "provider": "elasticsearch",
        "config": {
            "es_url": "http://localhost:9200",
            "index_name": "memories",
        }
    }
}

Dimension Mismatch Fix

For custom embedding models with non-standard dimensions (e.g., 768 instead of 1536), add "embedding_model_dims": 768 to the vector store config.