Deploy DeepSeek-R1 on VPS: Installation Guide for AI Model Hosting

Deploy DeepSeek-R1 on VPS, a powerful reasoning-based Large Language Model (LLM), on a VPS allows you to leverage its AI capabilities remotely and efficiently. Unlike traditional LLMs, DeepSeek-R1 is designed for complex reasoning tasks in coding, mathematics, and science. This guide covers how to install and run DeepSeek-R1 on your VPS, ensuring an optimized, scalable, and secure deployment.

Why Run DeepSeek-R1 on a VPS?

A VPS offers dedicated resources, making it an ideal environment for running AI models without relying on local hardware. Here’s why you should consider setting up DeepSeek-R1 on a VPS:

• Scalability: Easily upgrade RAM, storage, or CPU as needed.
• Remote Access: Run AI models from anywhere with internet access.
• Reduced Hardware Requirements: No need for an expensive local GPU setup.
• Performance & Stability: VPS servers are optimized for high-load tasks.

Prerequisites

Before installing DeepSeek-R1, ensure your VPS meets the following minimum requirements:

• Operating System: Ubuntu 20.04+ (recommended) or Debian-based Linux distribution.
• CPU: Multi-core processor (Intel i9 or AMD Ryzen recommended).
• RAM: At least 16GB RAM (higher for larger models).
• Storage: Minimum 40GB SSD (DeepSeek-R1 models require significant disk space).
• GPU (Optional): NVIDIA GPU with CUDA support for faster processing.
• Python 3.8+ installed
• Ollama Installed (for easy model deployment)

Step 1: Setting Up Your VPS

Connect to Your VPS

To begin, connect to your VPS using SSH. Open a terminal and run:

ssh user@your-vps-ip

Replace user with your VPS username and your-vps-ip with your server’s IP address.

Update & Upgrade System Packages

Before proceeding with installation, update your system to the latest packages:

sudo apt update && sudo apt upgrade -y

This ensures a stable and secure software environment.

Step 2: Install Ollama

Download Ollama

Ollama is a tool that simplifies the deployment of LLMs like DeepSeek-R1. To install Ollama, run the following command:

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

Selecting the Correct Version

After installation, verify by running:

ollama

If installed correctly, the command will display the Ollama version and options.

Step 3: Download & Install DeepSeek-R1 Model

Installing Ollama Setup

Once Ollama is installed, download the DeepSeek-R1 model:

ollama pull deepseek-r1:14b

This command downloads the 14-billion-parameter version of DeepSeek-R1. Depending on your VPS specifications, you can choose a smaller or larger model.

Running the Installer

Step 4: Running DeepSeek-R1 on Your VPS

Installation Progress

After downloading, you can run the model by executing:

ollama run deepseek-r1:14b

This starts the DeepSeek-R1 model, allowing you to interact with it directly.

Verifying Installation

Step 5: Creating a Python Script to Use DeepSeek-R1

For programmatic interaction, use Python to communicate with DeepSeek-R1. First, install the Ollama Python library:

pip install ollama

Create a Python script, deepseek_chat.py, and add the following code:

import ollama
desired_model = 'deepseek-r1:14b'
question = 'What is the capital of France?'
response = ollama.chat(model=desired_model, messages=[
    {'role': 'user', 'content': question},
])
print(response['message']['content'])
Run the script:
python deepseek_chat.py

This sends a question to DeepSeek-R1 and returns the AI-generated response.

Running DeepSeek-R1 from Command Line

Step 6: Running DeepSeek-R1 as a Background Service

To keep DeepSeek-R1 running persistently, use screen or tmux.

Using screen

screen -S deepseek_session
ollama run deepseek-r1:14b

Press Ctrl + A, then D to detach the session. To reconnect, use:

screen -r deepseek_session

Using tmux

tmux new -s deepseek_session
ollama run deepseek-r1:14b

Detach using Ctrl + B, then D. To reconnect, use:

tmux attach -t deepseek_session

Step 7: Securing & Optimizing DeepSeek-R1 on VPS

Selecting the DeepSeek-R1 Model

Enable Firewall

To prevent unauthorized access, enable a firewall:

sudo ufw allow ssh
sudo ufw allow 5000/tcp
sudo ufw enable

Monitor Performance

Monitor resource usage using:

top

gpu-smi (if GPU is available)

Choosing Model Parameters

Conclusion

Deploying DeepSeek-R1 on a VPS allows for efficient remote AI processing. By following these steps, you can install, configure, and run DeepSeek-R1 securely on your VPS. Whether for machine learning applications, research, or business, this setup ensures scalability and optimal performance.

For more advanced configurations, consider integrating Docker, optimizing for GPU acceleration, and implementing an API for remote access.By following this guide, you now have DeepSeek-R1 up and running on your VPS, ready for AI-driven applications!