Qdrant

Qdrant is a high-performance vector database and search engine with 35,000+ stars on GitHub, built in Rust for the next generation of AI applications.

Overview

Every serious AI product needs to search by meaning, not just keywords — and that is exactly the problem Qdrant solves. It stores embeddings and runs massive-scale nearest-neighbor search with an HNSW index, filtering and hybrid retrieval that combines dense vectors with sparse keyword signals. Written in Rust, it is fast enough for production recommendation, image search, neural search and RAG workloads, and it exposes both REST and gRPC APIs with official clients for popular languages. Qdrant runs anywhere from a laptop to a distributed cluster, and the same engine powers Qdrant Cloud, the fully managed service. The project is extremely active, with updates in October 2026, and it has become one of the default choices for vector search infrastructure.

Qdrant has also become a default backend for the AI ecosystem itself: LangChain, LlamaIndex, Open WebUI and dozens of other tools in this category ship Qdrant integrations, and its local mode — an in-process or single-Docker deployment with the same API as the cluster — makes prototyping trivial before you scale. Payload indexes keep filtered queries fast even across hundreds of millions of vectors, and features like multivectors and late interaction support modern retrieval techniques such as ColBERT-style ranking. Few open-source databases have matched its combination of speed, operability and API polish.

Key Features

• Vector search engine – HNSW-based approximate nearest-neighbor search at massive scale.
• Hybrid search – combine dense embeddings with sparse vectors and full payload filtering.
• Built for AI – embeddings similarity, neural search, image search and recommenders out of the box.
• Developer friendly – REST and gRPC APIs plus official SDKs for Python, Rust, Go, .NET and more.
• Scales out – sharding, replication and quantization for large, cost-efficient deployments.
• Managed cloud option – Qdrant Cloud runs the same engine as a service, with a free tier.

💳 Pricing

Qdrant is free and open source under the Apache License 2.0 for self-hosting. Qdrant Cloud offers a free tier and paid managed plans — see qdrant.tech for current pricing.

Who Is It For

Engineers building RAG systems, semantic or image search, recommendation engines and any application that needs fast, filtered vector retrieval in production.