What it is
Qdrant is an open-source vector search engine written in Rust and built in Germany. It stores embeddings alongside extra data (payload) and lets you filter on it within the same search, which matters for RAG and recommendations. Run it free with Docker, use Qdrant Cloud, or deploy the managed service inside your own cloud with its hybrid option.
How to use it
- Start Qdrant with a single Docker container or create a free cluster on Qdrant Cloud.
- Filter while you search by city, price or customer without losing speed.
- Mix dense and sparse vectors for hybrid search.
- Cut memory use with quantization as your collection grows.
- Connect it to your stack from Python, TypeScript, Rust or Go, or over REST and gRPC.
What it's good for
- A real estate portal recommending similar listings filtered by neighborhood and price.
- A team using n8n to build an agent with memory and search over internal docs.
- A fintech that must keep semantic search on its own infrastructure for compliance.
Pros
- Open source (Apache 2.0) and written in Rust
- Payload filtering built into vector search
- Free 1 GB cluster on Qdrant Cloud with no expiration date
- Hybrid Cloud option runs the managed service on your own infrastructure
Cons
- Self-hosting in production means handling replicas, backups and upgrades
- Tuning indexes and quantization takes some understanding of the internals
- Console and docs are English-only
