Local Quench

Download Local Quench

Free, no account. The app on your computer runs the agent and the model; the phone app and Safari connect to it.

  • You needA model server: LM Studio, llama.cpp (llama-server), Ollama or vLLM, or an API key for a cloud model.
  • Graphics card8 GB of VRAM for 9B models; 16 to 24 GB, or 8 GB plus 32 GB of system RAM, for 30B mixture-of-experts models. What runs on your card
  • UpdatesThe app checks for new versions and updates itself; the phone app updates from your computer.

    After installing

    Three steps to the first task.

    1. Load a model in LM Studio (or start llama-server). Local Quench finds it on its own; for a cloud model, add the key in Settings, Providers.
    2. Open a project folder and type /init: it writes the project memory, and reads CLAUDE.md or AGENTS.md if the project has them.
    3. Ask for a change. The agent reads, edits, runs the tests and asks before it touches anything outside the folder.

    On Android

    The app is not on Google Play yet. Android asks once to allow installing apps from your browser, and Play Protect may say the app comes from an unknown developer: tap More details, then Install anyway. Phone and tablet setup