---
title: "Deploy Ollama Self-Hosted (Docker)"
description: "Step-by-step guide to self-hosting Ollama with Docker Compose. "
---

# Deploy Ollama

Get up and running with Llama 3, Mistral, Gemma, and other large language models locally.

<div className="deploy-hero">
  <span className="deploy-hero-item">⭐ 60.0k stars</span>
  <span className="deploy-hero-item">📜 MIT License</span>
  <span className="deploy-hero-item">🔴 Advanced</span>
  <span className="deploy-hero-item">⏱ ~20 minutes</span>
  
</div>

<div className="mt-8 mb-4">
  <a 
    href="https://m.do.co/c/2ed27757a361" 
    target="_blank" 
    rel="noopener noreferrer"
    className="flex items-center justify-center w-full px-6 py-4 text-lg font-bold text-white transition-all bg-blue-600 rounded-xl hover:bg-blue-700 hover:scale-[1.02] shadow-lg shadow-blue-500/30"
  >
    🚀 Deploy on DigitalOcean ($200 Free Credit)
  </a>
</div>


## What You'll Get

A fully working Ollama instance running on your server. Your data stays on your hardware — no third-party access, no usage limits, no surprise invoices.

## Prerequisites

- A server with Docker and Docker Compose installed ([setup guide](/quick-start/choosing-a-server))
- A domain name pointed to your server (optional but recommended)
- Basic terminal access (SSH)

## The Config

Create a directory for Ollama and add this `docker-compose.yml`:

```yaml
# -------------------------------------------------------------------------
# 🚀 Created and distributed by The AltStack
# 🌍 https://thealtstack.com
# -------------------------------------------------------------------------

# Docker Compose for Ollama
version: '3.8'

services:
  ollama:
    image: ollama/ollama:latest # Official image is highly recommended for GPU support
    container_name: ollama
    ports:
      - "11434:11434"
    volumes:
      - ollama_data:/root/.ollama
    # For GPU support (NVIDIA), uncomment the following:
    # deploy:
    #   resources:
    #     reservations:
    #       devices:
    #         - driver: nvidia
    #           count: all
    #           capabilities: [gpu]
    networks:
      - ollama_net
    healthcheck:
      test: [ "CMD", "curl", "-f", "http://localhost:11434/api/tags" ]
      interval: 10s
      timeout: 5s
      retries: 5

networks:
  ollama_net:
    driver: bridge

volumes:
  ollama_data:
    name: ollama_data
```

## Let's Ship It

```bash
# Create a directory
mkdir -p /opt/ollama && cd /opt/ollama

# Create the docker-compose.yml (paste the config above)
nano docker-compose.yml

# Pull images and start
docker compose up -d

# Watch the logs
docker compose logs -f
```



## Post-Deployment Checklist

- [ ] Service is accessible on the configured port
- [ ] Admin account created (if applicable)
- [ ] Reverse proxy configured ([Caddy guide](/concepts/reverse-proxies))
- [ ] SSL/HTTPS working
- [ ] Backup script set up ([backup guide](/concepts/backups))
- [ ] Uptime monitor added ([Uptime Kuma](/deploy/uptime-kuma))

## The "I Broke It" Section

**Container won't start?**
```bash
docker compose logs ollama | tail -50
```

**Port already in use?**
```bash
# Find what's using the port
lsof -i :PORT_NUMBER
```

**Need to start fresh?**
```bash
docker compose down -v  # ⚠️ This deletes volumes/data!
docker compose up -d
```

## Going Further

- [Ollama on AltStack Directory](https://thealtstack.com/alternative-to/ollama)
- [Ollama Self-Hosted Guide](https://thealtstack.com/self-hosted/ollama)
- [Official Documentation](https://ollama.com)
- [GitHub Repository](https://github.com/ollama/ollama)