Loading…
14
Question about picking between a cloud AI API and running a local model on my own PC
Six months ago I finally sat down and picked between paying for a cloud AI API and running an open model on my own machine at home in Tucson. I went with the cloud one first because setup takes like five minutes and I didn't want to mess around with drivers at 1am. It worked great for a while, but then my bill hit $41 in one month just from messing around with little test projects. So I bought a used GPU with 12GB of memory and tried a local model instead, and honestly the answers were worse for anything with lots of steps. But it feels nice knowing my prompts never leave my desk, and I can run stuff at 3am without watching a meter spin. The thing I didn't expect is how much slower it feels when I'm waiting on my own card versus someone else's server. For anyone who has done both, did you end up staying local or going back to an API once the novelty wore off?
0 comments
Log in to join the discussion
Log In0 Comments
No comments yet
Be the first to share your thoughts on this discussion.