In our basement, an old gaming PC with an Nvidia GeForce RTX 3060 with 8 GiB VRAM (which breaks down when used for 3D rendering) runs a quantized GLM-4.7 on llama-agent. It's great for getting a feel for what marketeers call "Agentic AI" without exposing my data (or my thinking) to big-tech.
I get mixed results, but it was really useful for getting CI/CT of my old hobby project on GitHub back on its feet. I wouldn't have started doing that without using that system.
#llama-agent #homelab








