How to Run DeepSeek AI Locally on Your NAS – A Step-by-Step Guide

The world of AI is evolving rapidly, and DeepSeek AI is gaining significant attention for its ability to run locally with minimal hardware requirements. However, many users are concerned about data security and privacy when using cloud-based AI solutions. The best way to ensure your AI-powered workflows remain secure is to run DeepSeek AI on your own NAS. In this guide, we’ll walk you through setting up DeepSeek AI on a Synology NAS using Docker.

Affiliate links:
Synology DS420+ (Celeron CPU) – Struggled with AI models https://amzn.to/40SBd5r
Synology DS1621+ (Ryzen V1500B CPU) – Much better performance! https://amzn.to/4htqwxr

Why Run DeepSeek AI on a NAS?

Hosting AI locally on a NAS (Network Attached Storage) has several advantages:

• Data Security – Keep all your queries and interactions private.
• Offline Access – No need for an internet connection.
• Custom AI Workflows – Integrate with your personal documents, notes, and databases.
• Resource Efficiency – Run AI models without using a high-end workstation.

With Docker, installing and managing DeepSeek AI on a Synology NAS (or any NAS with Docker support) is straightforward. Below is a step-by-step guide on how to get it running.

Prerequisites

Before you begin, ensure you have the following:

✔️ A Synology NAS (Recommended: DS1621+ or better for smooth AI performance) ✔️ Docker installed (Available in Synology Package Center) ✔️ At least 8GB RAM (16GB+ preferred for faster performance) ✔️ Basic knowledge of using SSH (for troubleshooting if needed) ✔️ A volume with at least 10GB free space for AI model storage

Method 1: Deploy Using Docker Compose (SSH & CLI Method)

Step 1: Install Docker on Your NAS

1. Log in to your Synology DSM.
2. Open Package Center and search for Docker.
3. Install and open Docker.
4. Navigate to Registry and verify that Docker is working properly.

Step 2: Prepare DeepSeek AI Docker Setup

We will use Docker Compose to simplify the deployment. This will create two services:

1️⃣ DeepSeek Server – The core AI model handling requests. 2️⃣ WebUI – A user-friendly interface to interact with DeepSeek AI.

Creating the YAML Configuration File

1. Open File Station and create a directory for DeepSeek:

   /volume1/docker/deepseek/

2. Inside the deepseek folder, create subdirectories:

   webui
   data

3. Open SSH (Control Panel > Terminal & SNMP > Enable SSH Service).
4. Connect to your NAS via SSH:

   ssh admin@your-nas-ip

5. Create a docker-compose.yml file:

   cd /volume1/docker/deepseek
   nano docker-compose.yml

6. Copy and paste the following YAML configuration:

version: "3.9"

services:
  webui:
    container_name: DEEPSEEK-WEBUI
    image: ghcr.io/open-webui/open-webui:latest
    volumes:
      - /volume1/docker/deepseek/webui:/app/backend/data:rw
    environment:
      OLLAMA_BASE_URL: http://deepseek:11434  # Corrected IP for Ollama server
      DISABLE_AUTH: "true"  # No login required
    healthcheck:
      test: ["CMD", "bash", "-c", "timeout 10s bash -c ':> /dev/tcp/127.0.0.1/8080' || exit 1"]
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 90s
    ports:
      - "8271:8080"  # Access WebUI via http://your-nas-ip:8271
    restart: on-failure
    depends_on:
      deepseek:
        condition: service_started

  deepseek:
    container_name: DEEPSEEK-SERVER
    image: ollama/ollama:latest
    volumes:
      - /volume1/docker/deepseek/data:/root/.ollama:rw
    environment:
      MODELS: deepseek-r1:1.5b  # Change to 7B if needed
      OLLAMA_INSTALL_MODELS: deepseek-r1:1.5b
    ports:
      - "11434:11434"
    healthcheck:
      test: ["CMD", "ollama", "--version"]
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 30s
    restart: on-failure

Step 3: Deploy DeepSeek AI on Your NAS

1. Navigate to the deepseek directory:

   cd /volume1/docker/deepseek

2. Start DeepSeek AI using Docker Compose:

   docker-compose up -d

3. Verify that the containers are running:

   docker ps

   You should see DEEPSEEK-WEBUI and DEEPSEEK-SERVER running.

Method 2: Deploy Using Synology DSM Container Manager

If you prefer a graphical approach without SSH, use Synology’s Container Manager (Projects feature):

1. Open Container Manager in DSM.
2. Click on Projects in the left menu.
3. Click Create and give your project a name (e.g., DeepSeekAI).
4. Upload the YAML file (docker-compose.yml) or paste the script in the editor.
5. Click Next and configure the storage paths if necessary.
6. Set environment variables if needed (match the YAML configuration above).
7. Click Deploy and wait for the containers to start.
8. Once running, access DeepSeek WebUI at:

   http://your-nas-ip:8271

This method makes it easy to manage and restart the containers from DSM without using the command line.

Final Thoughts: Is Running DeepSeek AI on NAS Worth It?

Running DeepSeek AI on a NAS provides an excellent self-hosted, private, and secure AI solution. However, performance depends on the hardware:

✅ Good Performance: NAS with Ryzen (DS1621+, DS1821+) ⚠️ Slow Performance: Celeron-based NAS (DS420+ struggles with DeepSeek AI) 🚀 Best Performance: NAS with a GPU (e.g., Synology with PCIe support)

If you need faster processing, consider upgrading your NAS or adding a dedicated GPU.

💡 Would you like a guide on how to add a GPU to your NAS for AI processing? Let us know in the comments!

📢 Stay tuned for more AI and NAS tutorials on NASCompares!

Where to Buy a Product
			VISIT RETAILER ➤
			VISIT RETAILER ➤
			VISIT RETAILER ➤
			VISIT RETAILER ➤
			VISIT RETAILER ➤

☕ WE LOVE COFFEE ☕

Or support us by using our affiliate links on Amazon UK and Amazon US