Agreed. I have an additional question. Given what look like limitations on tasks of larger scope, what problems are people who claim that they have run the LLMs for days working on?