Qdrant

Qdrant is an Apache-licensed vector search engine by Qdrant that stores dense and sparse vectors, filters by payload, supports hybrid retrieval and runs self-hosted or as managed Qdrant Cloud.

Written by

What Qdrant does

Qdrant is a vector search engine for storing embeddings and finding similar records. Its documentation says collections support dense or sparse vectors, with named vectors letting one point hold different embedding types, and that payload indexes enable filtering during search. Qdrant describes hybrid retrieval that combines semantic and lexical search, and says it exposes HTTP and gRPC interfaces with official client libraries for Python, JavaScript and TypeScript, Rust, Go, .NET and Java.

Qdrant's deployment docs cover four ways to run it: self-hosted, Managed Cloud, Hybrid Cloud and Private Cloud. Qdrant Cloud is a managed service that runs the same engine as the self-hosted version on AWS, GCP and Azure, and Qdrant says you can pull the Docker image or clone the repo to run it yourself.

Who Qdrant is for

  • Teams that want an open source vector engine and the option of a managed cloud
  • Search that mixes vector similarity with payload filters, or dense with sparse retrieval
  • Agents that need Qdrant-backed semantic memory through an official MCP server

Who should pick something else

  • Always-on prototypes on the free cluster, because Qdrant suspends an unused free cluster after one week and deletes it after four weeks unless it is reactivated
  • Teams that need an uptime SLA on the free cluster: Qdrant's docs list uptime SLAs as a Standard cluster feature, not for free clusters

Qdrant: one strength, one limit

Strength. Qdrant's Cloud documentation says a free cluster, with 1 GB of RAM, 0.5 vCPU and 4 GB of disk, supports roughly 1 million vectors of 768 dimensions, and that no credit card is needed.

Limit. An unused free cluster is suspended after one week and deleted after four weeks unless it is reactivated.

What Qdrant costs

The engine is open source under Apache 2.0 and can be self-hosted. Qdrant Cloud has a free single-node cluster, a usage-based Standard tier and a Premium tier with a minimum spend; Hybrid Cloud and Private Cloud are other options.

Current prices: qdrant.tech . Prices change often, so check the maker's own page before you buy.

Qdrant with AI coding agents

Qdrant documents three official MCP servers: Qdrant Cloud MCP (cluster management, hosted by Qdrant, OAuth sign-in), Qdrant Docs MCP (documentation search, hosted, no account) and an open source Qdrant MCP server for semantic memory. It also provides Agent Skills, an llms.txt file and a Qdrant Cloud CLI (checked ).

We also list this company's other pages: Qdrant MCP server.

Questions people ask

Can Qdrant run on infrastructure I control?

Yes. Qdrant's Cloud page says you can pull the Docker image or clone the repo to run the same engine yourself, and Qdrant also lists Hybrid Cloud and Private Cloud deployment models.

Does Qdrant have an MCP server?

Yes. Qdrant documents three official MCP servers: Qdrant Cloud MCP, Qdrant Docs MCP and an open source Qdrant MCP server for semantic memory and code search.

Alternatives to Qdrant

Compare Vector databases tools

All three have a free plan.

Qdrant uses the Apache-2.0 license; Weaviate uses the BSD-3-Clause license; Pinecone is proprietary.

Compare all 3

More in Vector databases