ModelGo v0.1.0-alpha

ModelGo v0.1.0-alpha.1

First public test release of ModelGo, an Android application for downloading and running quantized GGUF language models locally.

Features

• Runs GGUF models locally using llama.cpp
• Supports Android 10 and newer
• Supports 64-bit ARM devices (arm64-v8a)
• Curated selection of mobile-friendly Q4 quantized models
• Model download progress inside the application
• Android download notifications
• Automatic detection of downloaded models
• Local chat interface
• Model reset and unload controls
• New ModelGo launcher icon and startup screen

Tested models

• TinyLlama 1.1B Chat Q4_K_M
• Qwen2.5 1.5B Instruct Q4_K_M

Requirements

• Android 10 or newer
• 64-bit ARM Android device
• At least 4 GB RAM recommended
• Sufficient storage for GGUF model files

Known limitations

• Only arm64-v8a devices are supported
• Responses are displayed after generation completes; token streaming is not implemented
• Chat history is not permanently stored
• Performance depends heavily on the device and model
• Larger models may fail on low-memory devices
• This APK is currently signed with a development/debug key

Installation

1. Download the APK.
2. Allow installation from unknown sources when prompted.
3. Install the APK.
4. Open ModelGo and download a supported model.

This is an early testing release and is not production-ready.