Bluehost Self-Managed VPS: Reimage Your Server for DeepSeek Installation

DeepSeek now runs directly on Bluehost Self-Managed VPS Hosting using a quick one-click installation. This setup keeps your data completely private and lets you talk to the AI through a simple terminal or your own apps. You just choose your server size, click install, and start interacting with your new AI assistant immediately.

• Reimage Your Server for Deepseek Installation Using Bluehost Portal
• How to Test and Verify Your DeepSeek Setup
• Additional Information & Example Documentation Output
• Summary

Reimage Your Server for Deepseek Installation Using Bluehost Portal

To reimage your server and install Deepseek:

1. Click the Manage button on the Self-Managed VPS package.

   
    

2. Click the Reimage button.

   
    

3. Select the Applications tab.

   
    

4. From the list, find Deepseek and click Select.

   
    

5. To confirm, please type "reimage" and then click Proceed to start the process.

   
    

6. Wait a few moments while the installation completes.

   
    

   • Once completed, you will see DeepSeek listed in the Server Image section.

     
      

How to Test and Verify Your DeepSeek Setup

Here is how to quickly check your server, run DeepSeek commands, and make sure everything is working safely.

1. Log in to your server via SSH.

   When you log in to your server for the first time via SSH, you're greeted with a welcome message that includes all the essential commands and instructions for verifying DeepSeek. The content looks like this:

   ssh [email protected]

   Example Output

   
   Welcome to Ubuntu 24.04.4 LTS (GNU/Linux 6.8.0-124-generic x86_64)
   ********************************************************************************
   DeepSeek AI Server
   API : http://123.45.67.89:11434 (Ollama-compatible REST)
   SSH : port 22 | UFW enabled - all other ports blocked
   Default model : deepseek-r1:1.5b (~1.1 GB, auto-pulled on first boot)
   Requirements:
   RAM : 4 GB minimum for this image (default model: deepseek-r1:1.5b)
   Larger models need more - see table below
   Model RAM requirements (minimum node RAM / recommended):
   deepseek-r1:1.5b -> 4 GB min / 8 GB recommended (default)
   deepseek-r1:7b -> 8 GB min / 16 GB recommended
   deepseek-r1:14b -> 16 GB min / 24 GB recommended
   Quick start:
   ollama run deepseek-r1:1.5b # interactive chat (default, needs 4 GB RAM)
   ollama list # list loaded models
   ollama pull deepseek-r1:7b # larger model (~4.7 GB, needs 8 GB RAM)
   ollama pull deepseek-r1:14b # pull an even larger model (~9 GB, needs 16 GB RAM)
   REST API:
   curl http://123.45.67.89:11434/api/tags
   curl -X POST http://123.45.67.89:11434/api/generate \
   -H 'Content-Type: application/json' \
   -d '{"model":"deepseek-r1:1.5b","prompt":"Hello","stream":false}'
   Service : systemctl status ollama
   Logs : /var/log/deepseek/deepseek.log
   Docs : cat /root/README.md
   ********************************************************************************
   To delete this message of the day: rm -rf /etc/update-motd.d/99-deepseek
   Last login: Tue Jun 9 18:08:22 2026 from 180.190.20.22
   

2. Check the Health of the Ollama Service:

   
   Check if the service is active:
   systemctl status ollama
   

   Example Output

   
   root@server-123456:~# systemctl status ollama
   ● ollama.service - Ollama Service
        Loaded: loaded (/etc/systemd/system/ollama.service; enabled; preset: enabled)
       Drop-In: /etc/systemd/system/ollama.service.d
                └─logging.conf, override.conf
        Active: active (running) since Tue 2026-06-09 17:22:47 UTC; 54min ago
      Main PID: 763 (ollama)
         Tasks: 39 (limit: 19161)
        Memory: 15.0G (peak: 15.0G)
           CPU: 7min 58.247s
        CGroup: /system.slice/ollama.service
                ├─ 763 /usr/local/bin/ollama serve
                └─6179 /usr/local/lib/ollama/llama-server --model /usr/share/ollama/.ollama/models/blobs/sha256-aabd4debf0>
   

3. Run and Manage DeepSeek Models via CLI:

   ollama run deepseek-r1:1.5b

   Example Output

   
   root@server-123456:~# ollama run deepseek-r1:1.5b
   pulling manifest
   pulling aabd4debf0c8: 100% ▕██████████████████████████████████████████████████████████▏ 1.1 GB
   pulling c5ad996bda6e: 100% ▕██████████████████████████████████████████████████████████▏  556 B
   pulling 6e4c38e1172f: 100% ▕██████████████████████████████████████████████████████████▏ 1.1 KB
   pulling f4d24e9138dd: 100% ▕██████████████████████████████████████████████████████████▏  148 B
   pulling a85fe2a2e58e: 100% ▕██████████████████████████████████████████████████████████▏  487 B
   verifying sha256 digest
   writing manifest
   success
   >>> good morning
   
   Good morning! 🌞 How can I assist you today?
   
   >>> Send a message (/? for help)
   

4. Run the Test Suite:

   prove /root/app_test/main.t

   Example Output

   
   root@server-123456:~# prove /root/app_test/main.t             prove /root/app_test/main.t
   /root/app_test/main.t .. ok
   All tests successful.
   Files=1, Tests=7,  0 wallclock secs ( 0.01 usr  0.01 sys +  0.04 cusr  0.03 csys = 0.09 CPU)
   Result: PASS
   

Additional Information & Example Documentation Output

You can run cat /root/README.md to see more information and useful commands:

Example Output:

Docs    : cat /root/README.md
********************************************************************************
To delete this message of the day: rm -rf /etc/update-motd.d/99-deepseek
Last login: Tue Jun  9 21:26:30 2026 from 123.123.20.22
root@server-123456:~# cat /root/README.md
# DeepSeek

## Description

DeepSeek is an advanced AI model platform that lets you run powerful language model capabilities directly on your own server infrastructure.
It provides access to high-performance AI reasoning and text generation,
allowing you to build applications, automate workflows, and process information with full control over your data and environment.

## Minimum Requirements

### Node resources

| Resource | Minimum | Recommended |
|----------|---------|-------------|
| RAM      | 4 GB    | 8 GB (default model); see model table below for larger models |
| Disk     | 5 GB    | 20 GB+ (if pulling multiple or larger models) |
| GPU      | None (CPU-only supported) | NVIDIA (CUDA) or AMD (ROCm) GPU for faster inference |

> **GPU note:** Ollama automatically offloads model layers to any detected NVIDIA (CUDA)
> or AMD (ROCm) GPU. Without a GPU, inference runs on CPU only - functional but noticeably
> slower, especially for larger models. For production workloads or low-latency responses,
> a GPU-equipped node is strongly recommended.

### RAM requirements by model

RAM figures account for model weights (Q4_K_M quantization), KV cache at default context
length, Ollama runtime (~150 MB), and Ubuntu OS idle usage (~500 MB).

| Model              | Disk size | Min node RAM | Recommended RAM |
|--------------------|-----------|--------------|-----------------|
| deepseek-r1:1.5b   | ~1.1 GB   | 4 GB         | 8 GB            |
| deepseek-r1:7b     | ~4.7 GB   | 8 GB         | 16 GB           |
| deepseek-r1:14b    | ~9.0 GB   | 16 GB        | 24 GB           |
| deepseek-r1:32b    | ~20 GB    | 32 GB        | 32 GB           |
| deepseek-r1:70b    | ~43 GB    | 64 GB        | 64 GB           |

The default model pre-loaded on first boot is `deepseek-r1:1.5b`. Pulling a larger model
on a node with insufficient RAM will cause Ollama to return a `500 Internal Server Error`.

Following installation the virtual machine will have:
* DeepSeek REST API on port 11434 (Ollama-compatible, OpenAI-compatible)
* SSH on port 22

Ports are protected using ufw with default-deny incoming policy.

## Services

Service           | Port  | Protocol | Notes
------------------|-------|----------|------
SSH               | 22    | TCP      | UFW rate-limited
DeepSeek API      | 11434 | TCP      | Ollama runtime, OpenAI-compatible endpoint

## Links

* DeepSeek models on Ollama: https://ollama.com/library/deepseek-r1
* Ollama documentation: https://github.com/ollama/ollama
* Ollama REST API reference: https://github.com/ollama/ollama/blob/main/docs/api.md
* DeepSeek official site: https://www.deepseek.com

## How-to-use

### Interacting via CLI

SSH into the server and use the `ollama` command:

ollama list # list downloaded models
ollama run deepseek-r1:1.5b # interactive chat session (default, needs 4 GB RAM)
ollama pull deepseek-r1:7b # pull a larger model (~4.7 GB, needs 8 GB RAM)
ollama pull deepseek-r1:14b # pull an even larger model (~9 GB)
ollama rm deepseek-r1:1.5b # remove a model

### Interacting via REST API

The API is accessible from outside the server on port 11434:

# List available models
curl http://:11434/api/tags

# Generate a completion (non-streaming)
curl -X POST http://:11434/api/generate \
     -H 'Content-Type: application/json' \
     -d '{"model":"deepseek-r1:1.5b","prompt":"Explain quantum computing","stream":false}'

# Chat endpoint
curl -X POST http://:11434/api/chat \
     -H 'Content-Type: application/json' \
     -d '{"model":"deepseek-r1:1.5b","messages":[{"role":"user","content":"Hello"}]}'

# OpenAI-compatible endpoint
curl -X POST http://:11434/v1/chat/completions \
     -H 'Content-Type: application/json' \
     -d '{"model":"deepseek-r1:1.5b","messages":[{"role":"user","content":"Hello"}]}'

### Model storage

Models are stored in /usr/share/ollama/.ollama/models. Ensure sufficient disk space before pulling large models (7B models require ~4.7 GB, 14B ~9 GB, 32B ~20 GB).

### Service management

systemctl status  ollama       # check service status
systemctl restart ollama       # restart the service
journalctl -u ollama -f        # follow journal logs
tail -f /var/log/deepseek/deepseek.log   # follow file logs (rotated daily, 14 days)

### Running the test suite

prove /root/app_test/main.t

### Security note
The Ollama API on port 11434 is exposed to the network and does not provide authentication by default. Restrict access using cloud firewall/security groups, a reverse proxy with authentication, or deploy only on trusted private networks.

Summary

DeepSeek now runs directly on Bluehost Self-Managed VPS Hosting using a quick one-click installation. This setup keeps your data completely private and lets you talk to the AI through a simple terminal or your own apps. You just choose your server size, click install, and start interacting with your new AI assistant immediately.