🤖 LM Studio + OpenWebUI Integration

Mục đích: Scripts to connect LM Studio (local LLM) with Open WebUI for a ChatGPT-like interface
Platform: Windows, Mac, Linux
Components: LM Studio + Open WebUI

---

🎯 What This Is

LM Studio provides local LLM inference (run models like Llama, Mistral on your machine).
Open WebUI provides a beautiful ChatGPT-like web interface.

This repo contains connection scripts to bridge them together.

---

📁 Files

• start_lmstudio_connection.bat - Windows batch script
• start_lmstudio_connection.ps1 - Windows PowerShell script
• start_lmstudio_connection.sh - Mac/Linux shell script
• README_WINDOWS.md - Windows-specific setup guide

---

🚀 Quick Start

Prerequisites

1. LM Studio installed and running

   • Download: https://lmstudio.ai/
   • Load a model (e.g., Llama 3, Mistral 7B)
   • Start local server (default: http://localhost:1234)

2. Open WebUI installed

   • Docker method: docker run -p 3000:8080 ghcr.io/open-webui/open-webui:main
   • Or install locally: https://github.com/open-webui/open-webui

---

🔧 Usage

Windows

# Option 1: PowerShell (recommended)
.\start_lmstudio_connection.ps1

# Option 2: Batch file
.\start_lmstudio_connection.bat

Mac/Linux

chmod +x start_lmstudio_connection.sh
./start_lmstudio_connection.sh

---

🌐 Access

After starting:

1. LM Studio: http://localhost:1234 (LLM server)
2. Open WebUI: http://localhost:3000 (Chat interface)
3. Configure: In Open WebUI settings, set API endpoint to http://localhost:1234/v1

---

⚙️ Configuration

LM Studio Settings

• Port: 1234 (default)
• API Type: OpenAI-compatible
• CORS: Enable if needed

Open WebUI Settings

1. Go to Settings → Connections
2. Set "OpenAI API Base URL" to http://localhost:1234/v1
3. API Key: Not needed for local (use dummy key like lm-studio)

---

💡 Why Use This?

Benefits:

• ✅ Privacy: All data stays local (no cloud)
• ✅ Free: No API costs
• ✅ Fast: No internet latency
• ✅ Offline: Works without internet
• ✅ Customizable: Use any model you want

Use Cases:

• Private coding assistant
• Sensitive document analysis
• Learning/experimenting with LLMs
• Offline work environments

---

🔧 Troubleshooting

LM Studio not responding?

• Check if model is loaded
• Verify server is running (green light in LM Studio)
• Test: curl http://localhost:1234/v1/models

Open WebUI can't connect?

• Check API URL is exactly http://localhost:1234/v1
• Disable CORS if needed in LM Studio settings
• Check firewall/antivirus not blocking

Slow responses?

• Use smaller models (7B instead of 70B)
• Enable GPU acceleration in LM Studio
• Reduce context length

---

📚 Resources

• LM Studio Docs: https://lmstudio.ai/docs
• Open WebUI Docs: https://docs.openwebui.com/
• Model Hub: https://huggingface.co/models

---

🔄 Recommended Models

For coding:

• DeepSeek Coder 6.7B
• CodeLlama 7B/13B
• Phind CodeLlama 34B

For general chat:

• Llama 3 8B/70B
• Mistral 7B
• Mixtral 8x7B

Vietnamese support:

• Vistral 7B (Vietnamese-optimized)
• SeaLLM 7B (Southeast Asian languages)

---

Last Updated: 2025-10-13
Maintained By: Khogao
License: Private