Application hosting

Hosting Open WebUI

Open WebUI is a self-hosted interface for working with language models through connected inference backends. Its official setup paths include Docker and Kubernetes, with persistent application data and a secret key that must survive container replacement. The project does not publish a universal CPU or memory minimum.

Min vCPU
Depends on workload
Min memory
Depends on workload
Min storage
Depends on workload

Must have

Hard prerequisites

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.

Recommended

Documented guidance

No further general recommendations are published for this version.

Capacity depends on your workload

Open WebUI does not publish a general CPU, RAM, or disk minimum. Requirements depend on its bundled components, user count, retained data, and whether inference runs on the same machine. GPU capacity is not required when the inference backend is hosted separately.

We only compare what providers publish. A RAM figure alone can't confirm compatibility or production capacity.

Deploy with

Supported paths

  • Docker container with a persistent volume
  • Docker Compose
  • Kubernetes using the project Helm chart

The host must provide

Runtime

Container runtime
Docker for the quick-start container path; Kubernetes for the Helm deployment path
Inference backend
Connect a supported remote or local model service; running inference locally has separate hardware requirements
Persistent secret
Set and retain WEBUI_SECRET_KEY across container recreation

Storage

Persistent data

Persist the application data directory at /app/backend/data. It holds application data such as users, settings, and chats. Size storage for retained conversations and any model or embedding assets stored with the chosen setup; the installation guide does not specify a universal disk size.

Step by step

Plan an Open WebUI deployment

  1. Choose where inference runs

    Decide whether models will run on this host or on a separate service. The WebUI container itself is an interface; local inference adds its own CPU, memory, and possibly GPU requirements.

    Official instructions
  2. Keep data and the signing key

    Mount persistent storage for /app/backend/data and set a strong WEBUI_SECRET_KEY that remains the same when updating or recreating the container.

    Official instructions
  3. Prepare for shared use

    Before exposing the service, configure authentication, persistent storage, backups, and monitoring. Use the project deployment guidance for a Kubernetes installation.

    Official instructions

Security

Before you go public

  • Do not expose an unauthenticated instance to the public internet.
  • Store WEBUI_SECRET_KEY in server-side secrets and keep it stable across deployments.
  • Back up the persistent data volume and restrict access to administrative accounts.

Cost

Beyond the server price

  • Open WebUI software does not include the compute cost of a separately hosted inference backend.
  • Budget for persistent storage, backups, and the host. A Kubernetes multi-replica deployment can require additional shared services described by the project documentation.

Matching plans

Plans with published evidence

DigitalOcean

CPU-Optimized Regular Droplet · 2 vCPU / 4 GiB

2vCPU
4 GiBRAM
25 GiBStorage
4 TBTransfer
  • Dedicated CPU
  • SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
DigitalOcean

General Purpose Droplet · 2 Dedicated vCPU / 8 GB

2vCPU
8 GBRAM
25 GBStorage
4 TBTransfer
  • Dedicated vCPU (100% allocatio...
  • NVMe SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Google Cloud

Compute Engine e2-medium · 2 vCPU / 4 GB

2vCPU
4 GBRAM
—Storage
—Transfer
  • Shared vCPU (1.00 sustained)
  • Persistent Disk (billed separa...
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Google Cloud

Compute Engine e2-small · 2 vCPU / 2 GB

2vCPU
2 GBRAM
—Storage
—Transfer
  • Shared vCPU (0.50 sustained)
  • Persistent Disk (billed separa...
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Hetzner Cloud

Hetzner Cloud CCX13 (Germany and Finland)

2vCPU
8 GBRAM
80 GBStorage
20 TBTransfer
  • Dedicated AMD vCPU
  • NVMe SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Hetzner Cloud

Hetzner Cloud CCX23 (Germany and Finland)

4vCPU
16 GBRAM
160 GBStorage
20 TBTransfer
  • Dedicated AMD vCPU
  • NVMe SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Hetzner Cloud

Hetzner Cloud CCX33 (Germany and Finland)

8vCPU
32 GBRAM
240 GBStorage
30 TBTransfer
  • Dedicated AMD vCPU
  • NVMe SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Hetzner Cloud

Hetzner Cloud CCX43 (Germany and Finland)

16vCPU
64 GBRAM
360 GBStorage
40 TBTransfer
  • Dedicated AMD vCPU
  • NVMe SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Hetzner Cloud

Hetzner Cloud CCX53 (Germany and Finland)

32vCPU
128 GBRAM
600 GBStorage
50 TBTransfer
  • Dedicated AMD vCPU
  • NVMe SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Hetzner Cloud

Hetzner Cloud CCX63 (Germany and Finland)

48vCPU
192 GBRAM
960 GBStorage
60 TBTransfer
  • Dedicated AMD vCPU
  • NVMe SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Hetzner Cloud

Hetzner Cloud CPX11 (US locations)

2vCPU
2 GBRAM
40 GBStorage
1 TBTransfer
  • Shared AMD vCPU
  • NVMe SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.
Hetzner Cloud

Hetzner Cloud CPX12 (Germany and Finland)

1vCPU
2 GBRAM
40 GBStorage
20 TBTransfer
  • Shared AMD vCPU
  • NVMe SSD
Compatibility unclear from published plan data

Meeting documented runtime prerequisites does not prove production capacity. Traffic, workloads, software configuration, and shared host limits are not measured here.

  • For the documented container path, run Docker; the quick-start supports Linux and macOS on x86_64 or ARM64, plus Windows through Docker Desktop.This prerequisite has no comparable published plan field.

Want the full monthly bill? Use the hosting cost calculator. It keeps unknown charges visible.