Llama Windows

Llama Windows puts llama.cpp, a widely used engine for running open AI models, in your Windows 11 tray. Pick a model from the Recommended Models list, download it in one click, and chat with it in a local web page. No terminal, no build step, no admin rights.

Each model shows its size on disk and whether it handles images, so you can check it fits first. Since v0.13.0 (Oct 20), fit badges say whether each model fits your PC's memory. Once a model is downloaded, everything runs on your PC with no internet needed. Builders get a local server at localhost:9931 that works with OpenAI-style code.

Windows 11 users who want to try a private, offline AI chat, and builders who want a local model their own scripts can call.

Skip it if you're on Windows 10, or you want ChatGPT- or Claude-level answers rather than open models (on a Mac, use Llama for Mac).

Llama Windows is free and open source (MIT license). Checked Oct 28, 2026 on the official pricing page.

• No setup pain. It installs llama.cpp itself, with no terminal or admin rights.
• Fully local. Prompts and data stay on your PC, and it works offline.
• Works with your code. Apps and agents can call its OpenAI-compatible server at localhost:9931.
• One model cache. Models land in the standard Hugging Face cache, shared with other tools.

• Windows 11 only. Windows 10 isn't supported; Mac and Linux users need the separate Llama for Mac app or command-line version.
• Open models only. It runs open models from the Hugging Face Hub, not ChatGPT, Claude or Gemini.
• Brand new. It launched October 13, 2026, so there's little track record yet.
• Gated models need a token. Some models require a Hugging Face access token in Settings.

Llama Windows, Ollama and LM Studio all run open models locally for free. Llama Windows is a Windows 11 tray app that sets up llama.cpp for you; the same team makes Llama for Mac and a Linux command-line version. Ollama and LM Studio also cover Mac and Linux.

Pick Llama Windows for the simplest start on a Windows 11 PC. Pick Ollama or LM Studio if you want one tool across all your systems.

1. On a Windows 11 PC, go to llama.app and download Llama Windows.
2. Launch it and find the llama icon in the notification area.
3. Pick a model under Recommended Models and download it.
4. Load the model and open the web UI to start chatting.
5. Point an OpenAI-compatible script at http://localhost:9931/v1 if you build.