Qdrant Reference

This section documents the current Qdrant 1.x line (latest 1.19.1), written and verified against the official Qdrant documentation. No specific patch version is pinned. Some capabilities (Qdrant Cloud-only features and the full breadth of client-library APIs beyond Python and Java) are linked, not documented in depth.

This content was generated with the assistance of AI and should be verified against the official documentation before being relied on in production.

This section’s bibliography lists the reference material consulted while preparing these pages.

Welcome to the Qdrant reference. Qdrant is an open-source, self-hostable vector similarity search engine written in Rust: it stores high-dimensional vectors alongside a JSON payload in collections, indexes them with an HNSW approximate-nearest-neighbor graph, and exposes both a REST and a gRPC API (with official Python and Java clients, among others) for filtered similarity search, hybrid dense/sparse retrieval, and the retrieval step of Retrieval-Augmented Generation pipelines.

If you are new to Qdrant, read Getting Started first, then Collections & Points and Vector Indexing & HNSW, followed by Filtering & Payload Indexing, Hybrid & Sparse Search, CRUD & Batch Operations, Clustering & Storage, Security & Administration, and finally RAG Integration Patterns. Everything after the first two pages builds on those foundations.

What’s covered

  • Getting Started — Qdrant as a vector similarity search engine written in Rust, running it via Docker or Qdrant Cloud, and a first create-collection/upsert/search round-trip over REST and the Python client.

  • Collections & Points — collections (fixed vector size and distance metric) and points (id + vector + payload), plus named vectors and multivectors.

  • Vector Indexing & HNSW — how Qdrant builds and tunes its HNSW index per collection or named vector, exact search, and scalar/product/binary quantization.

  • Filtering & Payload Indexing — payload indexes (keyword, integer, geo, text), the must/should/must_not boolean filter model, and filtered HNSW search.

  • Hybrid & Sparse Search — named sparse-vector configs, Reciprocal Rank Fusion via the Query API’s prefetch mechanism, and prefetch + cross-encoder re-ranking.

  • CRUD & Batch Operations — upserting, retrieving, and deleting points individually or in batches, and paging through a collection with the scroll API.

  • Clustering & Storage — sharding and replication, Raft-based cluster consensus, on-disk vs. in-memory storage trade-offs, and snapshots.

  • Security & Administration — static API keys, TLS, Qdrant Cloud role-based access control, and the /metrics and /telemetry endpoints.

  • RAG Integration Patterns — chunking and embedding a corpus, storing chunks as points with metadata payloads, and wiring retrieval up via LangChain or LlamaIndex.

  • Cheat Sheet (PDF) — a single-page, printable summary of everything in this section, with a downloadable PDF.

Bibliography