A small cheap VPS won’t have the compute or RAM to run these. The best way (and the intent) is to run it locally. A fast box with at least 32GiB of RAM (or VRAM

A small cheap VPS won’t have the compute or RAM to run these. The best way (and the intent) is to run it locally. A fast box with at least 32GiB of RAM (or VRAM for a GPU) can run many of the models that work with llama.cpp. For this 40G model you will need more like 48GiB of RAM. Apple Silicon is pretty good for local models due to the unified CPU/GPU memory but a gaming PC is probably the most cost effective option. If you want to just play around and don’t have a box big enough then temporarily renting one at Hetzner or OVH is pretty cost effective.