Llama Windows: llama.cpp in Your Windows Tray
Hugging Face says this is the Windows counterpart to llama.app on macOS: a native WinUI3 tray app that downloads a model from the Hugging Face Hub in one click and runs it locally, with no terminal, no build step and no admin rights, with native support for Windows on Arm and the new NVIDIA RTX Spark
Tool Overview
Details, capabilities and usage guidance
Features
- ✓ A free and open source tray app: Hugging Face says Llama Windows is a free and open source Windows 11 app that puts llama.cpp in your system tray. Pick a model from the Hugging Face Hub, download it in one click and run it locally from the notification area - no terminal, no build step, no admin rights
- ✓ Native WinUI3 and Fluent Design: Hugging Face says it is a native WinUI3 app that embraces Microsoft Fluent Design and lives in the notification area. Click the llama icon and a compact flyout opens right above it; there is no taskbar window and nothing else to manage, and right-clicking the icon opens or exits the app
- ✓ One-click model downloads from the Hub: Hugging Face says you can browse recommended Hugging Face models and download them in one click, replacing the old process of grabbing the right binary for your hardware, finding a model, choosing between quantised builds and launching a server with the right flags
- ✓ Native support for Windows on Arm and RTX Spark: Hugging Face says the app runs natively on Windows on Arm and that it worked with NVIDIA to have Llama Windows ready for RTX Spark on day one, with llama.cpp running models on the RTX Spark GPU. Hugging Face says that because llama.cpp is a highly optimised runtime you get full use of RTX Spark's compute and memory without manual tuning
- ✓ Data stays on your PC, and it can serve a local API: Hugging Face says everything runs on device so your prompts and data stay on your PC. Third-party coverage notes that it exposes a local OpenAI-compatible server that your own scripts and agents can use. The code is open source under the ggml-org organisation and available at llama.app
How to Use
- Download Llama Windows from llama.app on any Windows 11 PC (the code is at github.com/ggml-org/Llama-Windows)
- After installing, click the llama icon in the notification area and browse and one-click download a model from the Hugging Face Hub in the flyout
- Run it locally and chat with the model - Hugging Face says there is no terminal, no build step and no admin rights needed, and your data stays on your PC
- If you want your own scripts or agents to call it, check the community notes about the local OpenAI-compatible server it exposes, and treat the official repository docs as the source of truth
FAQ
What is Llama Windows?
Hugging Face says Llama Windows is a free and open source app that puts llama.cpp in your Windows 11 tray: pick a model from the Hugging Face Hub, download it in one click and run it locally from the notification area, with no terminal, no build step and no admin rights.
How does it relate to llama.app on macOS?
Hugging Face says Llama Windows is the Windows counterpart to llama.app on macOS, and that it launches with native support for Windows on Arm and for the new NVIDIA RTX Spark.
Do I need a specific GPU?
Hugging Face says the app runs natively on Windows on Arm and that it worked with NVIDIA to be ready for RTX Spark on day one, with llama.cpp running models on the RTX Spark GPU. Third-party coverage notes it also runs on normal x64 Windows 11 PCs - an RTX Spark is not required.
Where do I download it, and is the code open?
Hugging Face says you can download it from llama.app on any Windows 11 PC, or browse the code at its GitHub repository, ggml-org/Llama-Windows. Hugging Face says it is a free and open source app, and the repository builds for both x64 and ARM64.
Is my data uploaded anywhere?
Hugging Face says everything runs on device, so your prompts and data stay on your PC. Hugging Face frames this as combining llama.cpp's performance with a native Windows experience without sacrificing privacy.
Can my own programs call it?
Third-party coverage says Llama Windows exposes a local OpenAI-compatible server your own scripts and agents can use. The official blog focuses on the tray app experience, so check the official repository docs for interface details.