Skip to content
GhostRDP
  • AI
  • Linux VPS
  • Ubuntu 24.04

Run Ollama and Open WebUI on a VPS

Ollama with Open WebUI runs small AI models on CPU on a GhostRDP Linux VPS Pro plan with 6 vCPU and 12 GB RAM in Germany or the Netherlands, from €25.99/mo.

Ollama runs open language models on your own server and Open WebUI adds a chat interface in the browser, so prompts never leave the server. Our servers have no GPU, so models run on the CPU: small models (1B to 3B) answer at a usable speed, 7B and 8B models work but slower.

Order Pro · €25.99/moInstall steps

  • Chat in your browser

    Open WebUI gives Ollama a ChatGPT-style interface.

  • Private prompts

    Models run on your server; nothing is sent elsewhere.

  • Runs on CPU

    Small and mid-size open models, no GPU needed.

Plans for Ollama and Open WebUI

Memory decides which models fit. Pro runs 7B and 8B models with room for Open WebUI; Standard is enough for 1B to 3B models. More cores make answers faster.

  • Linux VPS Standard

    €12.99/mo

    €11.04/mo billed yearly, save €23/year

    • 4 vCPU
    • 8 GB ECC RAM
    • 70 GB NVMe
    • 1 Gbps port
    Order Standard

  • Linux VPS Pro

    Recommended

    €25.99/mo

    €22.09/mo billed yearly, save €46/year

    • 6 vCPU
    • 12 GB ECC RAM
    • 90 GB NVMe
    • 1 Gbps port
    Order Pro

  • Linux VPS Advanced

    €34.99/mo

    €29.74/mo billed yearly, save €62/year

    • 12 vCPU
    • 24 GB ECC RAM
    • 150 GB NVMe
    • 1 Gbps port
    Order Advanced

What you need

  • Operating system
    Linux VPS with Ubuntu 24.04
  • Memory
    About 4 GB RAM for 3B models, 8 GB or more for 7B and 8B models, plus about 1 GB for Open WebUI
  • GPU
    None. Ollama runs in CPU-only mode on our servers

Official site: ollama.com

How to install Ollama and Open WebUI

Written for Ubuntu 24.04 from the official instructions. Run the commands as root and replace YOUR_SERVER_IP with your server's address.

  1. 1

    Connect to your server over SSH

    Use the IP address and root password from your welcome email (the server page has a Copy SSH command button). Windows 10 and 11, macOS and Linux all include an SSH client in the terminal.

    Shell
    ssh root@YOUR_SERVER_IP
  2. 2

    Update the system

    Install the latest security updates before you add anything else.

    Shell
    apt update && apt -y upgrade
  3. 3

    Install Ollama

    Ollama's official install script sets it up as a service. It will note that no GPU was found and use the CPU.

    Shell
    curl -fsSL https://ollama.com/install.sh | sh
systemctl status ollama
  4. 4

    Download a model

    Downloads Llama 3.2 3B (about 2 GB) and asks it a question to check that it works.

    Shell
    ollama run llama3.2:3b "Explain what a VPS is in one sentence."
  5. 5

    Install Docker

    Docker's official convenience script adds Docker's own package repository and installs Docker Engine with the Compose plugin.

    Shell
    curl -fsSL https://get.docker.com -o get-docker.sh
sh get-docker.sh
docker --version
  6. 6

    Start Open WebUI

    The command from Open WebUI's README for Ollama on the same server. It uses the host network so it can reach Ollama on 127.0.0.1, and serves the interface on port 8080.

    Shell
    docker run -d --network=host \
  -v open-webui:/app/backend/data \
  -e OLLAMA_BASE_URL=http://127.0.0.1:11434 \
  --name open-webui --restart always \
  ghcr.io/open-webui/open-webui:main
  7. 7

    Create your admin account

    Open http://YOUR_SERVER_IP:8080 straight away and create your account; the first account becomes the administrator. Pick llama3.2:3b in the model menu and start chatting.

After installing

  • Models
    Add models with ollama pull <name> (browse ollama.com/library) or from Open WebUI's admin settings. ollama rm <name> frees disk space.
  • Ports
    8080 for Open WebUI. Ollama's API (11434) stays on localhost. For HTTPS, put Open WebUI behind a reverse proxy such as Caddy with your domain.
  • Speed
    On CPU, expect seconds rather than milliseconds per answer. Small models and short prompts are fastest.

Ollama and Open WebUI questions

Which plan do I need for Ollama and Open WebUI?

Pro for 7B and 8B models, Standard for 1B to 3B models. Memory is what limits model size.

Do your servers have a GPU?

No. Ollama runs on the CPU, which suits small models and private, low-volume use.

Are my prompts private?

The models run on your own server, and Ollama and Open WebUI store your chats there. Nothing is sent to an outside AI service unless you add one.

Can I use Ollama without the web interface?

Yes. Skip the Open WebUI steps and use the ollama command or its API on port 11434, for example through an SSH tunnel (ssh -L 11434:localhost:11434 root@YOUR_SERVER_IP).

Can other people use my Open WebUI?

Yes. As the administrator you decide who can sign up and which models they see in the admin settings.

Stuck on a step? Our team answers on live chat and in tickets.

Related apps

More on GhostRDP

Run Ollama and Open WebUI on your own server.

Online in minutes. Pay with crypto, no ID needed.

Choose a plan