Qdrant

Overview

Qdrant is a vector similarity search engine and vector database written in Rust. It provides a production-ready service with an API to store, search, and manage points, which are vectors with an additional payload. Qdrant is tailored for extended filtering support, making it useful for neural-network or semantic-based matching, faceted search, and other applications.

Key Features

• Dense, sparse, and multi vector search, including late interaction models such as ColBERT.
• Filtering on JSON payload with keyword matching, full-text search, numeric ranges, geo-locations, and should, must, and must_not clauses.
• Hybrid search combining multiple vectors in a single query with configurable fusion strategies like Reciprocal Rank Fusion and Distribution-Based Score Fusion.
• Vector quantization and on-disk storage to reduce RAM usage and tune the trade-off between search speed and precision.
• Distributed deployment with sharding and replication, plus zero-downtime collection updates and resizing.
• REST and gRPC APIs, official client libraries for multiple languages, and a Web UI for exploring collections and monitoring deployment health.

Use Cases

• Semantic text search that finds meaningful connections in short texts beyond keyword-based search.
• Similar image search, such as visual food discovery where users find meals by appearance rather than names or ingredients.
• Extreme classification, including e-commerce product categorization with millions of labels.
• Recommendation, discovery, and faceted search across neural-network or semantic-based matching applications.

Getting Started

• Run the Qdrant container locally with docker run -p 6333:6333 qdrant/qdrant.
• Connect to the server with a client, for example Python using qdrant_client.QdrantClient with url http://localhost:6333\.
• Use the Quick Start Guide, detailed documentation, and the Qdrant Essentials course for further setup.

Deployment & Requirements

• The local Docker command starts an insecure deployment without authentication, open to all network interfaces, so the instance should be secured.
• Before deploying Qdrant to production, review the installation and security guides.
• Qdrant is also available as a fully managed Qdrant Cloud service, including a free tier.

Before You Adopt

• License: Apache-2.0. Review its terms before using, modifying, or distributing the project.
• The default local container deployment is insecure: it lacks authentication and is open to all network interfaces.