Running local large language models: not as difficult as you might think