Hugging Face launches Llama Windows, a one-click llama.cpp app for Windows 11

Hugging Face's new app lowers the barrier to running local models on Windows. Users browse the Hub, download a model in one click and run it from the system tray, with no command line needed.

The Hub also now tags llama.cpp GGUF metadata automatically when a repository is created, which makes quantized builds easier to find. It hosts Meta's Muse Spark Open weights, and the platform has passed 3 million public checkpoints.

The launch rides the local-AI enthusiasm visible this week, from a jailbroken PS5 running Qwen 3.8 27B at 20 tokens per second to a Hugging Face livestream on speculative-decoding speedups. Windows has lagged the Mac for consumer local-LLM tooling, and LM Studio and Ollama are the incumbents.

Watch adoption and GPU-backend support.