Skip to the content
Know your tech before you buy it. Makers: get listed from $49/mo

Qdrant

By Qdrant · qdrant.tech

  • Free plan
  • Windows
  • macOS
  • Linux
  • Web
  • API
  • Self-hosted
8.0 Club score Visit qdrant.tech

What Qdrant is

Qdrant is a vector similarity search engine with an API for storing, searching and managing vectors alongside payload data. Points can pair vectors with JSON metadata that helps filter results. Qdrant supports dense, sparse and multivector configurations, along with hybrid retrieval combining semantic and lexical search. These capabilities can support semantic search and recommendation systems. It can run locally using a Docker image or through Qdrant Cloud. The project provides REST and gRPC APIs, official client libraries for Python, JavaScript/TypeScript, Rust, Go, .NET and Java, and integrations including LangChain, LlamaIndex, Airbyte, Unstructured, DocArray and AutoGen. Its free cloud tier is intended for testing and prototypes and is limited to a single-node cluster with 1GB RAM and 4GB disk. Qdrant Cloud includes TLS and encryption at rest; self-hosted deployments require configuration for those controls. Authentication and encryption are not enabled by default in self-hosted open-source deployments. The maker is based in Berlin and was founded in 2021.

Who it suits

Qdrant suits developers building semantic search or recommendation systems with vector and metadata filtering. Teams can choose local or self-hosted use, or Qdrant Cloud, depending on their deployment needs.

Good to know, and what to watch

  • Supports dense, sparse and multivector configurations.
  • Hybrid retrieval combines semantic and lexical search.
  • Offers REST and gRPC APIs and multiple client libraries.
  • Can run locally or through Qdrant Cloud.
  • Free cloud tier is limited to one node.
  • Free cloud tier has 1GB RAM and 4GB disk.
  • Self-hosted open-source deployments lack default authentication and encryption.
  • Self-hosted security controls require customer configuration.

Free Tier

Free

Hybrid Cloud

—

Premium Tier

—

Private Cloud

—

Free plan

Getting started

  1. Open the Qdrant website
  2. Choose Qdrant Cloud or a self-hosted deployment
  3. Run the Docker image for a local installation
  4. Connect through REST or gRPC APIs
  5. Use an official client library or listed integration
  6. Start with the free cloud tier for testing or prototypes

Limits to check

The Free Tier costs nothing and is billed free forever. It provides a single-node cluster with 0.5 vCPU, 1GB RAM, 4GB disk, and free cloud inference with selected models; it is intended for testing and prototypes.

The Club's review

Qdrant is a vector similarity search engine for applications that store embeddings with structured metadata. It suits developers building semantic search, recommendations, or hybrid retrieval who want both a Docker-based local deployment and a managed cloud option. Its broad vector support is a strength; self-hosted security and the small free cloud footprint require careful planning.

Overview

Qdrant stores, searches, and manages vectors through an API, with JSON payloads attached to each point. Those payloads make metadata filtering part of the retrieval workflow rather than a separate database step. That is a practical fit for search and recommendation systems where similarity results must also respect attributes such as category or access rules.

The project comes from Qdrant, founded in 2021 and headquartered in Berlin. You can run it locally with Docker or use Qdrant Cloud, giving a team a path from an application-controlled environment to managed operations without changing the product category.

For broader comparisons, see Vector Databases, Search Databases, Embedded Databases, and Database Software.

Key features

Vector and index choices

Qdrant supports dense, sparse, and multivector configurations, covering common embedding designs as well as more specialized representations. HNSW, sparse-vector, payload, and filterable-HNSW indexes let teams choose structures for approximate search and filtered queries. A 65,535-dimension maximum leaves substantial headroom, although most workloads will not need anything close to it.

Filtering and hybrid retrieval

JSON payload metadata can filter search results, which is valuable when a nearest-neighbor result is only useful within a tenant, product group, language, or other application-defined boundary. Hybrid retrieval combines semantic and lexical search, making Qdrant a better fit than a purely vector-oriented engine when exact terms and meaning both matter.

APIs, SDKs, and integrations

REST and gRPC APIs are accompanied by official clients for Python, JavaScript/TypeScript, Rust, Go, .NET, and Java. Integrations include LangChain, LlamaIndex, Airbyte, Unstructured, DocArray, and AutoGen. This range reduces adapter work for teams already using those ecosystems, while the API split gives infrastructure teams a choice between broadly accessible HTTP and gRPC-oriented services.

Deployment and security

Local Docker deployment gives operators control over placement and operations; Qdrant Cloud removes much of that infrastructure burden. Cloud includes TLS and encryption at rest, and security features are enabled by default. Open-source self-hosted deployments do not enable authentication or encryption by default, so production operators must configure those controls themselves. Qdrant provides compliance documentation for SOC 2 Type 2 and HIPAA through its Trust Center.

Pricing

Free Tier

The Free Tier costs free, billed free forever. It provides a single-node cluster with 0.5 vCPU, 1GB RAM, and 4GB disk, plus free cloud inference with selected models. That is suitable for learning, prototypes, and low-stakes validation, not for a growing production workload: the single node and tight memory and disk limits leave little room for resilience or large collections.

Standard and Premium tiers

Standard Tier has custom pricing and includes a 99.5% uptime SLA. Premium Tier also has custom pricing and raises the SLA to 99.9%. These are the managed choices for teams that need a stronger availability commitment than the free environment; the published material does not give a price, so budgeting requires a direct quote.

Hybrid and Private Cloud

Hybrid Cloud is custom priced, runs on customer infrastructure, is managed through Qdrant Cloud, keeps data inside the customer network, and targets production-grade uptime. It suits organizations balancing managed operations with network or data-location requirements. Private Cloud is also custom priced and provides a dedicated isolated deployment, custom SLAs, full isolation, and support for air-gapped deployment. That is the clearest fit for strict isolation requirements, but it is not a self-service budget tier.

Community support is Discord-only. Paid support offers either 10 hours per day during business hours or 24/7 coverage, which matters when an operational team cannot rely on community response.

Platforms

Qdrant supports API, Linux, macOS, Windows, web use through Qdrant Cloud, and self-hosted deployment. Docker makes local setup portable, while REST, gRPC, and six official SDK language families cover most application stacks.

Who it's for

Choose Qdrant when semantic search or recommendations are central and you need payload filtering, sparse or multivector support, or semantic-plus-keyword retrieval. It is especially suitable for teams that want to begin locally and later use managed cloud infrastructure, or that need hybrid and private-cloud deployment models.

It is less suitable when you need a tiny embedded database with no vector-specific operations, or when your team cannot take responsibility for configuring authentication and encryption in a self-hosted installation. The free cloud tier is a testing environment rather than a dependable production foundation.

Pros and cons

Pros

  • Broad vector coverage: Dense, sparse, and multivector configurations support varied embedding strategies.
  • Useful constrained retrieval: JSON payload filtering and filterable indexes keep similarity results aligned with application metadata.
  • Hybrid search: Combining lexical and semantic signals helps workloads where exact wording still matters.
  • Deployment flexibility: Docker, Qdrant Cloud, Hybrid Cloud, and Private Cloud cover local, managed, network-contained, and isolated operating models.
  • Strong integration surface: REST, gRPC, six official SDK languages, and named framework integrations reduce connection work.

Cons

  • Self-hosted security is operator-owned: Authentication and encryption are off by default, creating mandatory production configuration work.
  • Free cloud capacity is narrow: One node, 1GB RAM, and 4GB disk constrain realistic data volumes and resilience.
  • Higher tiers require a quote: Standard, Premium, Hybrid Cloud, and Private Cloud have no published prices, making upfront comparison difficult.
  • Community support is limited to Discord: Teams needing formal response commitments must purchase a support option.

Alternatives

ObjectBox is a freemium database with a free core plan and Android, iOS, desktop, API, and self-hosted coverage; choose it when an embedded, cross-platform database is a closer match than vector retrieval.

SQLite is a free, no-license-required embedded SQL engine for any purpose; pick it when relational local storage matters more than vector search.

Chroma is an Apache 2.0-licensed open-source database with a cloud offering; consider it when that open-source and cloud combination fits your stack.

LadybugDB is a free MIT-licensed database with source code and pre-compiled binaries; choose it when that licensing model and database focus are preferable.

PouchDB is a free, open-source database for supported desktop, mobile, web, and self-hosted platforms; consider it for that broad client footprint.

RxDB offers a free core with replication and realtime sync; pick it when synchronized client data is the priority.

DuckDB UI is a free local browser-based SQL notebook with optional MotherDuck connection; choose it for interactive local analytics.

EJDB2 is a free MIT-licensed open-source database engine; consider it when a compact general-purpose engine is enough.

Verdict

Qdrant is a strong choice for teams building production-oriented semantic or recommendation search that need metadata filters, hybrid retrieval, and deployment flexibility. Its main reason to choose it is the combination of rich vector configurations and practical APIs across local and managed environments. Look elsewhere if you need an embedded general-purpose database, a larger free cloud footprint, or self-hosting with security enabled without operator work.

The verdict

Qdrant offers several vector configurations, hybrid retrieval and multiple deployment options. Plan security configuration for self-hosting, and note the free cloud tier’s single-node resource limits.

Questions people ask

Is Qdrant free?

Yes. The Free Tier is free and is billed free forever.

Which platforms can run Qdrant?

Qdrant is available through API, Linux, macOS, self-hosting, web, and Windows.

What APIs and SDK languages are provided?

It provides REST and gRPC APIs, with official libraries for Python, JavaScript/TypeScript, Rust, Go, .NET, and Java.

Can Qdrant run on my infrastructure?

Yes. You can run it locally with Docker, self-host it, or choose Hybrid Cloud or Private Cloud deployments.

Does Qdrant support hybrid search and filtering?

Yes. It supports hybrid semantic and lexical retrieval and filtering with JSON payload metadata.

What security does Qdrant Cloud provide?

Qdrant Cloud enables TLS and encryption at rest by default. Self-hosted deployments require configuration for those controls.

What the maker publishes

APIs and clients
Qdrant provides REST and gRPC APIs and official client libraries for Python, JavaScript/TypeScript, Rust, Go, .NET, and Java. source
Cloud security
Qdrant Cloud includes built-in TLS and encryption at rest, while self-hosted deployments require customer configuration for these controls. source
Compliance
Qdrant says its compliance documentation for SOC 2 Type 2 and HIPAA is available through its Trust Center. source
Data and filtering
Qdrant points can contain vectors and JSON payload metadata, which can be used to filter search results. source
Deployment
Qdrant can be run locally using its Docker image or used through Qdrant Cloud. source
Founded
2021 source
Free plan
Yes source
Free-tier limit
The free cloud tier is intended for testing and prototypes and is limited to a single-node cluster with 1GB RAM and 4GB disk. source
Headquarters
Berlin, Germany source
Hybrid retrieval
Qdrant supports hybrid retrieval combining semantic and lexical search. source
Integrations
Qdrant lists integrations including LangChain, LlamaIndex, Airbyte, Unstructured, DocArray, and AutoGen. source
Product
Qdrant is a vector similarity search engine with an API to store, search, and manage vectors with additional payload data. source
Search use cases
Qdrant describes vector databases as useful for semantic search and recommendation systems. source
Security configuration
Self-hosted open-source deployments do not enable authentication or encryption by default, while Qdrant Cloud enables security features by default. source
Support
The pricing page lists community support as Discord-only and paid support options with 10h/day business-hours or 24/7 coverage. source
Vector support
Qdrant supports dense, sparse, and multivector configurations. source

Checked against the maker's pages on 2 Oct 2026.