I'm just getting into it with ollama. Well, sad to say, it uses a lot of energy and doesn't help, but wastes my time. I tell it to do one thing and it doesn't, it wants context, I provide context, it claims it's something it's not. It's as if it wanted to halt my progress. Maybe, but idk it's useful at a code compilation helper, but I still have to find out how to use it with Jetbrains products and without having to use their account required AI product sending data to them.

For small things, like "write me a bash script to replace one string with another" it's fine.

But Deepseek R1 9GB for instance, wtf is it doing? I ask it to write a systemd unit file to run a single command and it starts talking about its thought process and half an hour later it still didn't come to a result, all the while my GPU cooler running at max.

Waste of time, money and energy. Overhyped