@garritfra It's even easier with Mozilla's llamafile. Just download a single binary with an embedded model and run it (also includes web interface).
llamafile.ai/