Skip to content
AgentsIndex

Qdrant

Rust vector database that pairs filtered, dense plus sparse and multivector search and can run at the edge.

Best for
AI teams who need fast, controllable vector search across production workloads.
GitHub stars
35.1k

What we checked

Pages last read
Facts checked
5 facts

From 5 sources

AI answers
Too few read

58 buyer-question answers read, November 2026

At a glance

Tags
Freemium, HIPAA, Open source

Code

Last push
15 October 2026

For developers

Works with
Prometheus, Grafana, Datadog, AWS, Google Cloud, Azure and 10 more

What is Qdrant?

Qdrant is a vector database for AI teams that handles similarity search with fast indexing, payload-aware filtering, and hybrid retrieval. Its Native Hybrid Search, Built-in Multivector, and Expansive Metadata Filters support production search and recommendation workloads, while the OpenAPI v3 spec and integrations with Kubernetes and Grafana simplify deployment and monitoring.

Customers include Telekom, Tripadvisor, OpenTable, Hubspot, and Canva. Plans include Free Tier, Standard Tier usage-based pricing, Premium Tier minimum spend required, Hybrid Cloud, and Private Cloud.

An open-source vector search engine with a managed cloud around it

Qdrant is a vector search engine written in Rust. Around it the company sells Qdrant Cloud (managed on AWS, GCP or Azure), Hybrid Cloud ("Bring your own Kubernetes with decoupled control and data planes") and Private Cloud for air-gapped setups, all listed on the homepage.

The retrieval features the homepage leads with are one-stage filtering ("Filters are applied during HNSW traversal"), native hybrid dense and sparse search with BM25, SPLADE++ and miniCOIL, multivector objects, reranking with score boosting, ColBERT style late interaction and MMR, and quantization the vendor says cuts memory "by up to 64x". Access is through REST, gRPC and official client libraries, plus a built in web UI. Qdrant Cloud can also generate text and image embeddings itself through "Native Cloud Inference".

The free cluster is permanent, and paid cloud has no printed rate

The pricing page has three managed tiers and none shows a dollar figure. The Free Tier is "Free forever": a single node cluster with "0.5 vCPU / 1GB RAM/ 4 GB Disk" and free cloud inference with selected models. The FAQ says it is limited to "1GB RAM and 4GB disk without high availability", and outgrowing it means moving to Standard.

Standard Tier reads only "Usage-based pricing". The FAQ explains the meter: "You're charged for compute (vCPU), memory (GB), storage (GB) consumed by your clusters, storage (GB) consumed by backups, and used inference tokens of paid models. Usage is billed hourly." Standard adds dedicated resources, horizontal and vertical scaling, high availability, backups and a 99.5% uptime SLA, and it can be bought through a cloud marketplace. No per vCPU or per GB rate appears on the page; the only way to see a cost is the pricing calculator, which needs your vector counts first.

Premium Tier says "Minimum spend required" without naming the minimum, and is sold through "Talk to Our Engineers". The On-Premise tab of the same page lists Hybrid Cloud and Private Cloud with no price at all, only a contact button. A support table on the page names Community, Standard, Premium and Enterprise support levels, with a price shown only for the free Community level. There is no monthly or annual billing toggle.

The memory server for AI agents is published as an example

It exposes two tools: qdrant-store, which saves text with optional metadata, and qdrant-find, which runs a semantic search. A QDRANT_READ_ONLY setting disables the store tool.

It is official, but it does not expose collection management, filters or hybrid queries; an agent that needs those calls the REST or gRPC API directly. The homepage separately offers a setup prompt for coding agents that points at qdrant.tech/llms.txt.

Founded in 2021, with a $50M Series B in March 2026

The about page says Zayarni (CEO) and Andrey Vasnetsov (CTO) started the project in 2021 after finding libraries such as FAISS fell short, and that the team has "grown to 100+ experts across 20+ countries".

As of November 2026 the about page shows "9k Members" while the homepage shows "60k+ Community Members".

It suits engineers who want control, less so buyers who need a quote

Qdrant fits teams building RAG, agent memory, semantic search or recommendation systems who want filtered and hybrid retrieval in one engine, and who value being able to move between self-hosted, managed and hybrid deployments on the same open source core. The pricing FAQ says an OSS deployment can be migrated to Qdrant Cloud with a provided migration tool.

It fits less well for a buyer who must know the monthly bill before signing up. Standard is metered hourly across five resources with no published unit rates, Premium has an undisclosed minimum spend, and Hybrid and Private Cloud are priced only on contact. Teams that need SSO or private networking on the managed service are on Premium by definition. Self-hosting the free engine removes the bill but leaves scaling, backups and upgrades to your own engineers; the comparison table on the pricing page notes the OSS build has "No resource usage safe guards" and "No automated shard rebalancing".

How much does Qdrant cost?

Prices excluding tax, as Qdrant publishes them.

  • Free Tier

    Free

    • Single Node Cluster
    • 0.5 vCPU / 1GB RAM/ 4 GB Disk.
    • Free Cloud Inference With Selected Models
  • Standard Tier

    Usage-based pricing

    • Dedicated Resources
    • Flexible Vertical and Horizontal Scaling
    • Highly Available Setups
    • Backup & Disaster Recovery
    • Free Tokens for Paid Inference Models
    • 99.5% Uptime SLA
    • For production workloads and scaling applications
  • Premium Tier

    Minimum spend required

    • SSO
    • Private VPC Links
    • 99.9% Uptime SLA
    • Extra Support
  • Hybrid Cloud

    Runs on

    Run managed Qdrant clusters on your own infrastructure using your compute, network and storage.

    • Best for:
    • Local Data Residency
    • Regulated Workloads
    • Operations in Your Own Cloud
    • Benefits:
    • Data Stays in Your Network
    • Fully Managed Through Qdrant Cloud
    • Production-Grade Uptime
  • Private Cloud

    Runs on

    Dedicated, isolated deployment for strict security or compliance needs.

    • Best for:
    • Large Enterprises
    • Sensitive Workloads
    • Air-Gapped Setups
    • Benefits:
    • Custom SLAs
    • Full Isolation
Qdrant's pricing page

Pricing page, captured

qdrant.tech/pricing, checked

All plans and limits

PlanPriceMain limits
Free Tier$0 a month4 GB Disk; Nodes: Single Node Cluster; 1GB RAM
Hybrid CloudPrice on requestNone stated
Premium TierPrice on requestNone stated
Private CloudPrice on requestNone stated
Standard TierPrice on requestNone stated

Read from qdrant.tech/pricing.

What we checked on Qdrant

YesPartlyNo

What it costs to startYes

What the price counts

Managed cloud is billed on resource usage: vCPU, memory, storage, backups and paid inference tokens, charged hourly; Premium carries a minimum spend and no hourly rate is printed on the pricing page.

Qdrant Cloud pricing is based on resource usage, if you have a bigger cluster, you pay for more.

qdrant.tech/pricingChecked

What it costs to startYes

Free to use

The managed Free Tier is free forever and does not require a credit card.

Free forever (No credit card required)

qdrant.tech/pricingChecked

How you build with itYes

Public API

The engine is used through documented REST and gRPC APIs plus official client libraries for Python, JavaScript and other languages, with a public API reference.

Start with a single API call — scale to advanced control over HNSW, hybrid fusion, reranking, and multi-vector retrieval, all via REST, gRPC, or…

qdrant.techChecked

How you build with itYes

MCP server

Publishes its own MCP server on GitHub

An official Qdrant Model Context Protocol (MCP) server implementation

github.com/qdrant/mcp-server-qdrantChecked by us,

Questions about Qdrant

How much does Qdrant cost?
Qdrant has a free plan. Its paid plans include Hybrid Cloud, Premium and Private Cloud. From its own pages, checked 12 November 2026.
Does Qdrant have an API or an MCP server?
Qdrant has an API. From qdrant.tech, checked 12 November 2026. Qdrant has an MCP server. We checked github.com/qdrant/mcp-server-qdrant ourselves on 15 November 2026.
Is Qdrant open source, and can you host it yourself?
Qdrant is open source. We checked github.com/qdrant/qdrant/blob/master/LICENSE ourselves on 15 November 2026.
What have we checked on Qdrant?
5 facts, under these headings: What it costs to start, how you build with it.
How do we check Qdrant?
We hold 5 facts on Qdrant, quoted from 5 pages. Facts last checked 15 November 2026. Price last checked 12 November 2026. Last change found 15 November 2026.

How we check every listing

Sources and changes

Last full check

What changed2

  • What it does
  • What it does

Repository

Licence
Apache-2.0
Stars
35,090
Last push
15 October 2026

Qdrant price history

Qdrant was free on both of our reads, from 2 March 2026 to 12 November 2026.

PricePlanFirst seenLast seenReadingsSource
FreeManaged Cloud2 March 202612 November 20262Archived copy, 2 March 2026

From our research on 12 November 2026 and an Internet Archive copy from 2 March 2026. A date is when we read the page, not when the price changed.

Similar in Agent tools and integrations

See all in Agent tools and integrations

From our research

All research

Is Qdrant yours?

Free, and the page then carries the verified owner mark. You confirm it with an email at qdrant.tech. Claim this listing

Report an error if a fact is wrong.