Qdrant

What is Qdrant?

Qdrant (pronounced "quadrant") is an AI-native, open-source vector similarity search engine and database. It provides a production-ready service with a convenient API to store, search, and manage points (vectors with additional payload). Written in Rust, Qdrant delivers exceptional speed and reliability even under high load.

Qdrant excels in these AI scenarios:

Architecture Overview

Qdrant operates in a client-server architecture, exposing both HTTP and gRPC interfaces for seamless integration with any programming language. Its Kubernetes-native design supports horizontal scaling, automatic load balancing, and fault tolerance.

Core Concepts

{
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "vector": [0.1, 0.2, 0.3, ...],  // Dense vector
  "payload": {
    "city": "London",
    "category": "tech",
    "price": 99.99,
    "tags": ["AI", "ML"]
  }
}

Vector Types

Qdrant supports multiple vector types to handle different search scenarios:

Deployment Options

Qdrant offers flexible deployment to fit your infrastructure needs:

docker run -p 6333:6333 \
  qdrant/qdrant

Performance Features

Getting Started

# Install: pip install qdrant-client

from qdrant_client import QdrantClient
from qdrant_client.models import Distance, VectorParams, PointStruct

# Connect to local instance or cloud
client = QdrantClient("localhost", port=6333)
# Or: client = QdrantClient(url="https://xxx.cloud.qdrant.io", api_key="your-key")

# Create collection
client.create_collection(
    collection_name="my_collection",
    vectors_config=VectorParams(size=1536, distance=Distance.COSINE)
)

# Upsert points
client.upsert(
    collection_name="my_collection",
    points=[
        PointStruct(
            id=1,
            vector=[0.1, 0.2, 0.3, ...],
            payload={"city": "London", "category": "tech"}
        ),
        PointStruct(
            id=2,
            vector=[0.4, 0.5, 0.6, ...],
            payload={"city": "Berlin", "category": "science"}
        )
    ]
)

# Search with filtering
results = client.search(
    collection_name="my_collection",
    query_vector=[0.1, 0.2, 0.3, ...],
    query_filter={
        "must": [{"key": "city", "match": {"value": "London"}}]
    },
    limit=5
)

# RAG with Qdrant + OpenAI
from openai import OpenAI
from qdrant_client import QdrantClient

openai = OpenAI()
qdrant = QdrantClient("localhost", port=6333)

def rag_query(question: str) -> str:
    # 1. Embed the question
    embedding = openai.embeddings.create(
        model="text-embedding-3-small",
        input=question
    ).data[0].embedding

    # 2. Search Qdrant for relevant context
    results = qdrant.search(
        collection_name="knowledge_base",
        query_vector=embedding,
        limit=3
    )
    context = "\n".join([r.payload["text"] for r in results])

    # 3. Generate answer with context
    response = openai.chat.completions.create(
        model="gpt-4",
        messages=[
            {"role": "system", "content": f"Answer based on:\n{context}"},
            {"role": "user", "content": question}
        ]
    )
    return response.choices[0].message.content

Advanced Filtering

Qdrant's payload filtering is applied during the vector search phase (not after), enabling efficient filtered queries even on large datasets.

Integrations

Qdrant vs Pinecone

Use Cases

Learn More

Related Topics

• RAG (Retrieval-Augmented Generation)
• Embeddings
• Pinecone Vector Database
• Chroma Vector Database
• LangChain
• LlamaIndex