It's not useful for actual work, but the fact it can be run at all shows that we're evolving towards enabling powerful LLMs to run locally.