I'm surprised that most people still seem to use cloud-based ai models instead of local models, especially with the pretty big increase in local AI efficiency even on lower end hardware that we've seen these past two years.
For all of the people still using online AI / Sub systems, is it purely a hardware problem or what?