OmniChat AI 🧠💻

A privacy-first VS Code extension to chat with any local LLM through Ollama—including the new OpenAI open-weight models (gpt-oss-120b and gpt-oss-20b). Get code help, brainstorm, and run advanced AI—all offline. Your code and queries never leave your machine.

• Works with Any Major Open LLM: Use DeepSeek, Gemma, Llama 3, or the newest OpenAI models (gpt-oss-120b, gpt-oss-20b) through Ollama.
• Real-Time AI Interaction: Get coding assistance, brainstorm ideas, and debug issues without leaving VS Code
• Privacy-First: All processing happens locally - your code and data never leave your machine
• Beautifully Formatted Responses: Clean markdown rendering for better readability
• Seamless Integration: Matches VS Code's themes for a consistent experience

Installation

Prerequisites

• Visual Studio Code
• Ollama installed and running locally

Quick Install

1. Install from VS Code Marketplace:

   • Open VS Code
   • Go to Extensions (Ctrl+Shift+X)
   • Search for "OmniChat AI"
   • Click Install

2. Pull your preferred LLM with Ollama:

   ollama pull gpt-oss-120b
   ollama pull gpt-oss-20b
   # Or any other model you prefer
   ollama pull deepseek-r1:1.5b
   ollama pull gemma:7b
   ollama pull llama3:8b

Usage

1. Start Ollama in your terminal:

   ollama serve

2. In VS Code:

   • Open Command Palette (Ctrl+Shift+P)
   • Run OmniChat: Start
   • A chat panel will open

3. Set your preferred model:

   • Open Command Palette (Ctrl+Shift+P)
   • Run OmniChat: Set Model
   • Enter your model (e.g. gpt-oss-120b, gpt-oss-20b, deepseek-r1:1.5b, etc.)

4. Start chatting!

   • Type your question and click "Ask"
   • Receive beautifully formatted responses

Models

OmniChat AI works with any model available through Ollama. Some popular options:

• gpt-oss-120b — Large, enterprise-grade open-weight model
• gpt-oss-20b — Great for laptops and edge computers
• deepseek-r1:1.5b - Fast, lightweight coding assistant
• deepseek-r1:8b - Good balance of speed and capability
• deepseek-r1:32b - Most capable DeepSeek model
• gemma:7b - Google's lightweight model
• llama3:8b - Meta's efficient model

Tip: All you need is a single command (e.g. ollama pull gpt-oss-120b) and you’re ready to go.

Credits

This project was inspired by Fireship's video on building VS Code extensions.

Special thanks to Ollama for enabling local AI capabilities.

License

This project is licensed under the MIT License.