Leneex - Hosting Provider
Qdrant

Deploy Qdrant on a VPS in One Click

Databases & data storageAI & LLM tools

Vector database engine for semantic search, RAG pipelines, and AI recommendation systems.

Deploy on a VPSAll apps
  • Ready in ~3 min
  • Free installation
  • We handle the setup
At a glance
Version
Latest (1.19)
Operating system
Ubuntu Server 24.04 LTS
Minimum RAM
1 GB
License
Apache 2.0
Addressing
IPv4, IPv6
GitHub
35,000
Deployment~3 min
Open source & self-hosted
VPS plans

Choose a VPS plan for Qdrant

Blazing-fast NVMe drives and powerful processors. Take a ready-made plan or build the configuration yourself.
Minex
Perfect for personal projects and small websites.
€3.20/mo
Save €0.80 per month
  • 1 vCPU core
  • 1 GB DDR5 RAM
  • 20 GB NVMe SSD
  • Full root access
  • 1 dedicated IPv4
  • 1 TB traffic
Basex
A great balance of power and affordable price.
€4.80/mo
Save €1.20 per month
  • 2 vCPU cores
  • 2 GB DDR5 RAM
  • 40 GB NVMe SSD
  • Full root access
  • 1 dedicated IPv4
  • 2 TB traffic
Most popular
Corex
For growing applications and businesses.
€7.20/mo
Save €1.80 per month
  • 4 vCPU cores
  • 4 GB DDR5 RAM
  • 50 GB NVMe SSD
  • Full root access
  • 1 dedicated IPv4
  • Unlimited traffic
Midex
Resources for high-load projects.
€9.60/mo
Save €2.40 per month
  • 4 vCPU cores
  • 6 GB DDR5 RAM
  • 60 GB NVMe SSD
  • Full root access
  • 1 dedicated IPv4
  • Unlimited traffic
Proex
Maximum power and scalability.
€12.00/mo
Save €3.00 per month
  • 6 vCPU cores
  • 8 GB DDR5 RAM
  • 70 GB NVMe SSD
  • Full root access
  • 1 dedicated IPv4
  • Unlimited traffic
Scalex
Resources for scaling.
€16.00/mo
Save €4.00 per month
  • 6 vCPU cores
  • 10 GB DDR5 RAM
  • 150 GB NVMe SSD
  • Full root access
  • 1 dedicated IPv4
  • Unlimited traffic
Maxex
A large reserve of power.
€44.00/mo
Save €11.00 per month
  • 8 vCPU cores
  • 16 GB DDR5 RAM
  • 200 GB NVMe SSD
  • Full root access
  • 1 dedicated IPv4
  • Unlimited traffic
Apex
The top configuration.
€79.99/mo
Save €20.00 per month
  • 16 vCPU cores
  • 32 GB DDR5 RAM
  • 400 GB NVMe SSD
  • Full root access
  • 1 dedicated IPv4
  • Unlimited traffic

What you can do with Qdrant

Qdrant is an open-source vector database engine written in Rust, built for storing and searching high-dimensional embeddings at scale. It's aimed at ML engineers and AI application teams who need fast approximate nearest neighbor search for semantic search, RAG pipelines, or recommendation logic — not general-purpose relational storage.

What sets it apart is native payload filtering combined with vector similarity in a single query, so metadata like category, date, or user ID narrows results without a second database call. Running Qdrant self hosted on your own VPS puts collection configs, embedding model choices, and data residency fully in your hands — control a shared cloud tier abstracts away.

Get started
The applications screen of the control panel with one-click install

Qdrant features

A server surrounded by app icons — applications running on a VPS

Qdrant ships as a single binary with REST and gRPC APIs, ready to plug into any embedding pipeline once you set up hosting for Qdrant.

01

HNSW-based vector search

Approximate nearest neighbor search across millions of points with a tunable precision/speed trade-off.

02

Payload filtering

Combine vector similarity with structured metadata filters in one query, no join required.

03

Named vectors & multitenancy

Store several vector types per point and isolate tenants inside a single collection.

04

Quantization

Scalar, product, and binary quantization to shrink memory footprint on large collections.

05

Snapshots

Built-in collection snapshotting for point-in-time backup and restore.

06

REST and gRPC APIs

A Qdrant vps docker deployment exposes both interfaces on separate ports for language-agnostic clients.

07

Distributed mode

Optional clustering with sharding and replication for horizontal scale.

08

Bundled Web UI

A console for browsing collections, running queries, and visualizing embeddings without extra tooling.

Requirements

Which server to run it on

Minimum requirements for Qdrant and the configuration worth starting with.
Minimum
  • CPU1 vCPU
  • Minimum RAM1 GB
  • Disk10 GB
Recommended
VPS Basex
  • CPU2 vCPU
  • Minimum RAM2 GB
  • Disk40 GB

Installation is free — you only pay for the server

Getting started

How to deploy Qdrant

Three steps and the app is running on your server.
Step 1
Choose a server

Pick a plan with enough RAM for your collection size — Qdrant keeps vector indexes in memory by default, so an undersized instance slows searches as your dataset grows.

Step 2
Pick the app

Select Qdrant in the marketplace and the platform handles Qdrant installation for you, so you don't need to install Qdrant on vps manually via Docker or a .deb package.

Step 3
Get your credentials

Once deployment finishes, grab the REST/gRPC endpoint and any generated API key from the panel to connect your first client library.

Official websiteqdrant.tech
GitHub repositorygithub.com
Documentationqdrant.tech
Why us

Why Host Qdrant on Leneex

Speed, protection and support you can rely on — on every plan, with no hidden limits.
Deployment speed
A Qdrant vps hosting template starts the official Rust binary in seconds, so you're upserting embeddings almost immediately instead of waiting on a compile-from-source setup.
Free installation
Deploy Qdrant in one click without paying a setup fee — useful when you're just testing an embedding model before committing to a production cluster.
DDoS protection
Qdrant's REST/gRPC endpoint is often called directly by production RAG apps and search backends, so exposing it publicly benefits from traffic filtering that a database engine itself doesn't provide.
Control and support
Root/IPMI/API access lets you tune HNSW parameters, memory-mapped storage, and quantization settings a managed cloud tier typically locks behind fixed presets; engineers are reachable if something lower in the stack misbehaves.
99.9% uptime with SLA compensation
When Qdrant backs a live chatbot or search feature, index downtime breaks the whole user-facing pipeline — not just a background job — which is where SLA-backed uptime actually pays off.
Flexible payment
Pay by card or crypto, or get an invoice as a legal entity — fits both a solo ML project and a company running Qdrant as core retrieval infrastructure.
What to choose

Qdrant: Self-Hosted vs Cloud — Which to Choose

We compare cost, control over your data and limits, so the choice is an informed one.
Qdrant Cloud
Self-hosted on LeneexRecommended
Resources and limitsManaged clusters on shared or dedicated tiers, sized in fixed RAM/vCPU steps you pay for regardless of actual loadDedicated RAM/CPU/disk, sized to your collection — no neighbor tenants competing for the same index memory
Control and accessCluster settings exposed through the cloud console; low-level engine flags are largely fixed by the managed tierFull root access — adjust HNSW ef/m, storage mmap thresholds, quantization, and network exposure directly
Feature limitsFree tier caps cluster size and count; some advanced configuration options are reserved for higher paid tiersFull feature set: distributed mode, snapshots, all quantization types, no cluster-count caps
ScalingScale by upgrading the managed cluster tier, billed per resource stepResize the VPS or add nodes for distributed mode on your own schedule
CostFree tier for small workloads, then usage-based pricing tied to cluster size and uptimeFixed monthly VPS price regardless of query volume or storage growth
Who it fitsTeams that want to skip infrastructure ops entirely and accept usage-based billingProduction RAG/search systems needing predictable cost and low-level tuning

Self-hosted makes sense once your embedding traffic is steady enough that a fixed VPS price beats usage-based billing, or once you need engine settings the managed console doesn't expose. Qdrant Cloud remains the faster path for prototyping before you know your real query volume.

Frequently Asked Questions about Qdrant on VPS

What people ask most about Qdrant and our servers.

  • Yes, Qdrant stores data on disk as regular files, so resizing the server keeps the storage volume intact; snapshotting a collection first is still recommended before any resize.

  • It depends on collection size and vector dimensionality, since Qdrant keeps indexes in RAM by default; a small tariff comfortably handles development and low-volume production, while multi-million-vector collections need more memory.

  • You do — the VPS is yours to manage, so upgrading the Docker image or binary and restarting the service is a manual step, not an automatic platform update.

  • Yes — once the REST/gRPC ports are reachable and secured with an API key, any client library or automation tool that speaks Qdrant's API can connect remotely.

What's preconfigured

One-click deployment runs the official qdrant/qdrant Docker image with data persisted to a storage volume, so collections survive container restarts. Two ports are exposed by default: 6333 for the REST API and bundled Web UI, and 6334 for gRPC.

First access

Unlike apps with a traditional admin login, Qdrant has no username/password screen — access is controlled entirely through the API key (if one is set) and the endpoint URL. Open the bundled Web UI dashboard by adding /dashboard to your server's address on port 6333; if an API key was generated, the panel is the place to find it, since Qdrant itself won't show it again after creation.

A note on running without authentication

If no API key is configured, anyone who can reach ports 6333/6334 can read and write your collections — this is fine for a private network but risky on a public IP. Setting QDRANT__SERVICE__API_KEY (or the equivalent environment variable exposed in the Leneex panel) before loading real data is worth doing early, and restarting the container after setting it is required for the key to take effect.

Deploy Qdrant in about 3 minutes

Choose a server — we do the rest: install, configure and send you the credentials.

Choose a serverAll apps