# How to Install and Run [LM Studio](https://lmstudio.ai/) on Ubuntu as a Headless Server

This guide explains how to install and run LM Studio in a headless Ubuntu environment.

---

## **Prerequisites**
Ensure you have an Ubuntu 24.10 server with sufficient resources for the models you plan to use. For testing, a machine with **2 GB RAM, 2 CPUs, and 60 GB SSD** is sufficient for smaller models.

---

## **Step 1: Install Required Packages**
Run the following command to install all necessary dependencies:
```bash
sudo apt install -y npm fuse fuse3 libfuse2 libatk1.0-0 libatk-bridge2.0-0 libcairo2 libgdk-pixbuf2.0-0 libgtk-3-0 libx11-6 libnss3 libasound2 libasound2t64 libcups2 xauth xvfb xfce4 xfce4-goodies
```

---

## **Step 2: Download LM Studio**
Download the LM Studio AppImage from the official site (https://lmstudio.ai/):
```bash
wget https://installers.lmstudio.ai/linux/x64/0.3.6-8/LM-Studio-0.3.6-8-x64.AppImage
```

Make the file executable:
```bash
chmod +x LM-Studio-0.3.6-8-x64.AppImage
```

---

## **Step 3: Run LM Studio with a Virtual Display**
Run LM Studio using `xvfb` to simulate a graphical environment:
```bash
xvfb-run ./LM-Studio-0.3.6-8-x64.AppImage --no-sandbox
```
This initializes LM Studio and prepares it for headless operation.

---

## **Step 4: Install the LM Studio CLI**
Install the LM Studio CLI tool (`lms`) using:
```bash
npx lmstudio install-cli
```

During installation, you’ll see the following prompt:
```
We are about to run the following commands to install the LM Studio CLI tool (lms):
    echo 'export PATH="$PATH:/root/.lmstudio/bin"' >> ~/.profile
    echo 'export PATH="$PATH:/root/.lmstudio/bin"' >> ~/.bashrc
```
Confirm by typing `Yes`. This will add the CLI tool to your PATH.

Open a new terminal session or run:
```bash
source ~/.bashrc
```

Verify the installation:
```bash
lms --help
```

---

## **Step 5: Start the LM Studio Server**
Start the server to enable API access:
```bash
lms server start
```

Check the server status:
```bash
lms status
```
You should see output indicating the server is running on port **1234**.

---

## **Step 6: Download and Load a Model**
List available models to download:
```bash
lms get
```
Select a lightweight model such as **Qwen2.5 Coder 3B Instruct** and confirm the download. Once downloaded, load the model:
```bash
lms load qwen2.5-coder-3b-instruct
```

Verify the loaded models:
```bash
lms ps
```
---

## **Step 7: Test the Model via API**
Send a request to the running server to test the model:
```bash
curl http://localhost:1234/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen2.5-coder-3b-instruct",
    "messages": [
      { "role": "user", "content": "Write a Python function to reverse a string." }
    ]
  }'
```

### Example Response:
```json
{
  "id": "chatcmpl-o2nubfsik7bykwbkdysbrd",
  "object": "chat.completion",
  "created": 1736694420,
  "model": "qwen2.5-coder-3b-instruct",
  "choices": [
    {
      "index": 0,
      "logprobs": null,
      "finish_reason": "stop",
      "message": {
        "role": "assistant",
        "content": "Here's a simple Python function to reverse a given string:\n\n```python\ndef reverse_string(s):\n    # Using slicing to reverse the string\n    return s[::-1]\n\n# Example usage\nreversed_s = reverse_string(\"hello\")\nprint(reversed_s)  # Output: \"olleh\"\n```"
      }
    }
  ],
  "usage": {
    "prompt_tokens": 38,
    "completion_tokens": 116,
    "total_tokens": 154
  },
  "system_fingerprint": "qwen2.5-coder-3b-instruct"
}
```

---

## **Step 8: Manage Models**
### List Installed Models:
```bash
lms ls
```
### Unload a Model:
```bash
lms unload <model-identifier>
```
### Stop the Server:
```bash
lms server stop
```

---

## **Conclusion**
You have successfully installed and configured LM Studio on a headless Ubuntu server. You can now explore various models and use the API for your projects. If you encounter issues, refer to the LM Studio documentation or seek assistance. 😊