Local LLM client with an unexpected mining side gig

What it does

FreedomGPT is an Electron-and-React desktop wrapper around llama.cpp that runs LLM inference locally on macOS and Windows. It keeps chat offline by executing models on the user’s machine through a familiar chat interface. The app also includes an optional XMRig miner, letting users “enable mining earnings” alongside local inference.

The interesting bit

The pairing of a privacy-focused offline LLM tool with a cryptocurrency miner is, at minimum, a curious design choice. Beyond that, the project is largely integration work: it marries a React chat frontend—credited to the Chatbot UI project—to llama.cpp, with users manually downloading and pointing the app to Liberty Edge model files.

Key highlights

• Fully offline inference via llama.cpp; prompts never leave the machine
• Electron-based desktop app for Mac and Windows
• React chat interface adapted from Chatbot UI
• Manual model management: users supply their own Liberty Edge weights and configure paths
• Optional baked-in XMRig miner for Monero mining earnings

Caveats

• Model setup is manual and not plug-and-play
• Linux instructions appear partially copy-pasted from macOS (the README tells Linux users to copy XMRig into miner/mac/fgptminer)
• README offers little detail on architecture, security boundaries, or how the miner is isolated

Verdict

A decent choice for developers wanting a desktop UI over llama.cpp without cloud dependencies, provided you’re comfortable sourcing models yourself. Others should pass if bundled crypto miners are a dealbreaker, or if you need a polished, miner-free local-LLM experience.