Install Ollama on VPS: Complete Setup Guide 2026

ChatGPT Plus costs $20/month. Claude Pro $18/month. But what if you could chat with AI unlimited – for $9/month fixed costs?

The math: A VPS (Virtual Private Server – your own server in the cloud) with 8 GB RAM costs from $9/month. On it runs Ollama with Llama 3.2 – a language model that can keep up with GPT-3.5 for most use cases. No API costs, no token limits, full data control.

In this guide, you'll install Ollama on your own server. Afterwards you'll have: Llama 3.2 as local AI, a ChatGPT-like interface, and sensitive data that never leaves your network.

Prerequisites

Before we start, you'll need:

• A VPS with at least 8 GB RAM (for Llama 3.2 8B). Don't have one? Servers from $9/month with 8+ GB RAM →
• Ubuntu 22.04 or 24.04 (Debian works too)
• SSH access to the server
• Optional: A domain for HTTPS access

Our VPS recommendations for Ollama:

Contabo offers the best RAM/price ratio (16 GB for €10), Hetzner has the fastest CPUs for smoother responses, Netcup is a good middle ground.

Step 1: Install Ollama

Installation is a one-liner. Connect via SSH and run:

curl -fsSL https://ollama.com/install.sh | sh

Verify installation:

ollama --version
# Output: ollama version 0.5.x

The script installs Ollama to /usr/local/bin/ and sets up a systemd service that starts automatically.

Step 2: Download your first model

Now let's download Llama 3.2 8B - the best starter model:

ollama pull llama3.2

Test the model:

ollama run llama3.2
>>> Write a haiku about server hosting

Tip: The first start loads the model into RAM - this takes 10-30 seconds. After that, Ollama responds instantly.

Step 3: Enable remote access

By default, Ollama only listens on localhost. For external access (e.g., n8n, Open WebUI on another server):

sudo mkdir -p /etc/systemd/system/ollama.service.d/
sudo nano /etc/systemd/system/ollama.service.d/override.conf

Contents of override.conf:

[Service]
Environment="OLLAMA_HOST=0.0.0.0"

Restart the service:

sudo systemctl daemon-reload
sudo systemctl restart ollama

Warning: Without authentication, your API is now public! In the next step, we'll secure it.

Step 4: Secure the API

Three options for secure remote access:

Option A: Firewall (simplest)

Allow only specific IPs:

sudo ufw allow from YOUR_IP to any port 11434
sudo ufw enable

Option B: Reverse Proxy with Basic Auth

Traefik or Nginx with password protection - see Step 6.

Option C: WireGuard VPN

Most secure option - Ollama stays on localhost, access only via VPN with WireGuard.

Step 5: Install Open WebUI (ChatGPT Interface)

Open WebUI gives you a slick chat interface - including chat history and user management:

Install Docker (if not already installed):

curl -fsSL https://get.docker.com | sh
sudo usermod -aG docker $USER

Start Open WebUI:

docker run -d \
  --name open-webui \
  --restart always \
  -p 3000:8080 \
  -v open-webui:/app/backend/data \
  -e OLLAMA_BASE_URL=http://host.docker.internal:11434 \
  --add-host=host.docker.internal:host-gateway \
  ghcr.io/open-webui/open-webui:main

Open WebUI is now running on port 3000. The first user to register automatically becomes admin.

Step 6: Set up HTTPS with Traefik

For secure access via a domain, we'll set up Traefik as a reverse proxy:

version: '3'

services:
  traefik:
    image: traefik:v3.0
    command:
      - --api.insecure=true
      - --providers.docker=true
      - --entrypoints.web.address=:80
      - --entrypoints.websecure.address=:443
      - --certificatesresolvers.letsencrypt.acme.httpchallenge=true
      - --certificatesresolvers.letsencrypt.acme.httpchallenge.entrypoint=web
      - [email protected]
      - --certificatesresolvers.letsencrypt.acme.storage=/letsencrypt/acme.json
    ports:
      - 80:80
      - 443:443
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - ./letsencrypt:/letsencrypt
    restart: always

  open-webui:
    image: ghcr.io/open-webui/open-webui:main
    volumes:
      - open-webui:/app/backend/data
    environment:
      - OLLAMA_BASE_URL=http://host.docker.internal:11434
    extra_hosts:
      - host.docker.internal:host-gateway
    labels:
      - traefik.enable=true
      - traefik.http.routers.webui.rule=Host(`chat.yourdomain.com`)
      - traefik.http.routers.webui.entrypoints=websecure
      - traefik.http.routers.webui.tls.certresolver=letsencrypt
    restart: always

volumes:
  open-webui:

Tip: Replace chat.yourdomain.com with your domain and make sure the DNS A record points to your server IP.

Bonus: Connect Ollama with n8n

n8n has a native Ollama node for free AI automations:

1. Add an 'Ollama' node to your workflow
2. Configure the Base URL: http://localhost:11434 (if n8n runs on the same server)
3. Select your model (e.g., llama3.2)
4. Connect with Chat Trigger for a chat interface or other triggers for automations

Use cases:

• Automatically summarize and categorize emails
• Analyze and prioritize support tickets
• Generate content for social media
• Convert documents into structured data

Performance Tuning

Get the most out of your setup:

More context length

By default, Ollama uses 2048 tokens context. For longer conversations:

ollama run llama3.2 --num-ctx 4096

Warm up model

Load the model into RAM at server start for instant responses:

curl http://localhost:11434/api/generate -d '{"model": "llama3.2", "prompt": "warmup", "stream": false}'

Monitoring

Monitor RAM usage during operation:

watch -n 1 'free -h'

Model recommendations by use case

Common problems

Ollama doesn't respond from remote

Check:

1. OLLAMA_HOST=0.0.0.0 set?
2. Port 11434 open in firewall?
3. systemctl restart ollama executed?

Out of Memory error

Model is too large for your RAM. Use a smaller model or quantized version (e.g., llama3.2:8b-q4_0).

Slow responses

Normal on CPU! For 8B models expect 5-15 tokens/second. For more speed: Larger VPS or GPU server.

Open WebUI can't find Ollama

Check OLLAMA_BASE_URL and whether --add-host=host.docker.internal:host-gateway is set.

Conclusion

You now have a complete Ollama server with web interface and HTTPS. Your data stays on your server, you pay no API fees, and you can chat with AI unlimited.

Cost comparison: ChatGPT Plus costs $240/year. A VPS with Ollama costs ~€108/year – and you can also run Vaultwarden as password manager or Immich for photos on it.

More Self-Hosting Guides

n8n Self-Hosting Guide

Docker Setup Schritt für Schritt

Lesen

Vaultwarden Setup

Bitwarden auf eigenem Server

Lesen