Video by Hugging Face via YouTube
Run local models in Pi with llama.cpp — fully private, offline, and with no per-token cost. In this tutorial I show you how to install llama.cpp, pick the right GGUF model and quantization for your hardware using Hugging Face’s hardware compatibility feature, and download, load, and use the model directly from Pi with the `/llama` command. By the end you’ll have Qwen3 8B running locally as a coding agent, with no prompts, code, or data ever leaving your machine.
—
🔗 *Links*
– llama.app (llama.cpp installer + model recommendations): https://llama.app
– Qwen3.8 27B on Hugging Face (GGUF): https://huggingface.co/Qwen/Qwen3.8-27B
– Pi llama.cpp documentation: https://pi.dev/docs/latest/llama-cpp
– llama.cpp repository: https://github.com/ggml-org/llama.cpp
—
👋 *Connect with me*
– My website: https://alejandro-ao.com/
– X (Twitter): https://x.com/_alejandroao
– LinkedIn: https://www.linkedin.com/in/alejandro-ao/
—
🤓 *Topics Covered*
– Run local GGUF models in Pi with llama.cpp
– Choose the right quantization for your hardware
– Download and load models with Pi’s `/llama` command
—
⏱️ *Timestamps*
0:00 Intro — why local models in Pi
0:59 Install llama.cpp
1:49 Choose and load a model (Qwen3 8B)
5:24 Outro — hybrid local/cloud workflows