Manage GGUF models, run llama.cpp, and chat with local AI from one interface. LlamaControl runs locally, with no account or API key required.
Get started
Download the installer for your platform and open the app. LlamaControl is free and does not require an account. Linux support is coming soon.
Download LlamaControlAdd a model
Already have GGUF files? In Settings, choose your models folder or let the app detect it automatically. You can also find and download a model from Hugging Face in the model catalog.
Browse GGUF modelsInstall the engine
Open Settings and run the llama.cpp installer. The app detects your hardware and selects a suitable build for your GPU or CPU. You do not need to choose CUDA or Vulkan manually.
Run a model locally
Select a model, click Start, then open the Chat tab when the server is ready. If the model does not start reliably, Auto-Tune can test safer settings.
Use a coding agent
In Plugins, generate a configuration for OpenCode or Crush and paste it into your tool. The agent will connect to LlamaControl’s local server.
Next steps
Compare speed and quality in Benchmark, review your history in Analytics, or enable Remote Access to chat over your local network.