What Qdrant does
Qdrant is a vector search engine available as open-source software and managed cloud, supporting similarity search, payload filters, sparse vectors, quantization, and distributed clusters.
Qdrant separates the retrieval engine from the embedding and generation models. It supports dense and sparse vectors, payload filtering, multivectors, quantization, distributed replication, and APIs for search, recommendation, and hybrid retrieval. Teams can self-host the open-source engine or use Qdrant Cloud.
The free Cloud tier currently provides a single-node cluster with 0.5 vCPU, 1 GB RAM, and 4 GB disk for prototypes. Standard is usage-based with dedicated resources, scaling, backups, disaster recovery, and a 99.5% SLA. Premium requires a minimum spend and adds SSO, private VPC links, higher SLA, and support. Self-hosting avoids a managed subscription but adds infrastructure, security, backup, and on-call costs.
Deployment control also transfers responsibility. Public endpoints, broad API keys, missing TLS, untested snapshots, weak tenant filters, or stale replicas can expose data. Retrieval quality depends on the embedding model, distance metric, chunking, filters, and evaluation set. Enforce authentication and network boundaries, test recovery and deletes, benchmark recall, and keep citations and human review around consequential generated answers.
How Qdrant works
Clients create collections and insert points containing dense or sparse vectors plus payload metadata. Qdrant indexes the vectors, applies payload filters and similarity metrics, and returns nearest candidates; applications may fuse multiple searches or rerank results before generation. Operators choose Cloud or self-hosting and remain responsible for embeddings, permissions, backups, evaluation, and downstream output.
How to set up Qdrant
Choose managed or self-hosted
Compare compliance, region, networking, staff, availability, upgrade, backup, and total operating cost before selecting deployment.
Model collections deliberately
Select vector dimensions and distance, payload schemas, tenant strategy, sparse or multivector needs, provenance, and source version fields.
Secure and ingest a pilot
Require TLS and keys, restrict networks, minimize payloads, upsert representative records, and verify updates, deletes, snapshots, and restores.
Tune retrieval empirically
Measure recall, precision, latency, memory, filters, hybrid fusion, quantization, reranking, and isolation against labeled and adversarial queries.
Operate with change control
Monitor capacity and errors, rotate credentials, rehearse recovery, pin client compatibility, re-embed migrations safely, and audit permissions.
Qdrant FAQs
Is Qdrant open source?
Yes. The core vector database is open source, while Qdrant also sells managed Cloud and enterprise options.
Does Qdrant Cloud have a free tier?
Yes. It currently includes a small single-node cluster with 0.5 vCPU, 1 GB RAM, and 4 GB disk for testing.
What does Qdrant store?
It stores vector points and associated payload metadata; applications choose and operate the embedding and generation models.
Can Qdrant perform hybrid search?
Yes. Dense and sparse vector queries can be combined or fused, subject to collection design and client implementation.
Is self-hosting automatically more private?
No. It increases control but makes the operator responsible for network security, identity, encryption, patches, logs, backups, and incident response.
Listing reviewed 2026-07-24. Product details and pricing can change; verify important terms on the provider's website.
Related Research AI tools
Related AI guides
Reviews
Tell the community what you made, what worked, and what you wish you knew before starting.