Embex - The Universal Vector Database ORM

Embex is a vector database ORM that provides one API across Pinecone, Qdrant, LanceDB, Milvus, Chroma, Weaviate, and PgVector. It allows you to write your code once and run it with any supported provider.

The Problem

Every vector database currently has a different API. If you start with one provider and want to switch later—maybe for better performance, lower costs, or different features—you have to rewrite your entire codebase.

Without Embex

You’re locked into vendor-specific APIs:

Pinecone

# Pinecone specific code
index.upsert(vectors=[(id, values, metadata)])
results = index.query(vector=query, top_k=5)

Qdrant

# Qdrant specific code - completely different!
client.upsert(collection_name=name, points=points)
results = client.search(collection_name=name, query_vector=query, limit=5)

The Solution

Embex provides one unified API that works with seven different databases.

With Embex

Your core logic stays exactly the same. To switch providers, you only change one configuration line.

# Works with ANY provider
await client.collection("products").insert(vectors)
results = await client.collection("products").search(vector=query, top_k=5)

Switching from development (LanceDB) to production (Qdrant) is trivial:

client = await EmbexClient.new_async(provider="lancedb", url="./data")
client = await EmbexClient.new_async(provider="qdrant", url="http://localhost:6333")

Key Benefits

1. Unified API

Stop writing vendor-specific adapter code. Embex abstracts the differences between providers so you can focus on building your application.

2. High Performance

Embex isn’t just a wrapper. The core is written in Rust with SIMD acceleration, making vector operations up to 4x faster than native Python clients for tasks like normalization and cosine similarity.

3. Production Ready

Built for scale with features that many native clients miss:

• Connection Pooling: Efficiently manage database connections.
• Built-in Migrations: Manage schema changes across any provider.
• Observability: Hooks for monitoring and metrics.

Supported Providers

Embex currently supports:

• LanceDB: Perfect for local development and embedded apps (No Docker required).
• Qdrant: High-performance, scalable vector search engine.
• Pinecone: Fully managed vector database.
• Chroma: Open-source embedding database.
• Weaviate: AI-native vector database.
• PgVector: Vector similarity search for PostgreSQL.
• Milvus: Cloud-native vector database for massive scale.

Next Steps

📥 Installation Get Embex installed in seconds.

⚡ Quick Start Build your first vector app in 5 minutes.

📖 Tutorial Step-by-step guide to building a semantic search engine.