Very interesting. @curt.kennedy what is the typical response time for 1K total tokens when running on your local machine?