@matthewdechang/opencode-gemma4

Custom OpenCode MLX Provider for Google Gemma 4 (and other local MLX models).

Setup

This package provides a custom provider for OpenCode.ai to easily connect to your local MLX server (http://127.0.0.1:8083/v1).

1. Install

npm install -g @matthewdechang/opencode-gemma4

2. Configure opencode.json

Add the following to your .opencode.json or workspace configuration:

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "mlx": {
      "npm": "@matthewdechang/opencode-gemma4",
      "name": "MLX (local)",
      "models": {
        "google/gemma-4-26b-a4b-it": {
            "name": "Gemma 4 26B A4B (IT)"
        },
        "google/gemma-4-it": {
            "name": "Gemma 4 (IT)"
        },
        "mlx-community/Qwen2.5-Coder-7B-Instruct-8bit": {
            "name": "Qwen 2.5 Coder"
        },
        "mlx-community/NVIDIA-Nemotron-3-Nano-30B-A3B-4bit": {
            "name": "Nemotron 3 Nano"
        }
      }
    }
  }
}

Repository

https://github.com/mattthewdechang/opencode-gemma4