You are using the OpenRouter option, yes? You don't have to use it technically.
If you have a good enough GPU for your computer, at minimum 8GB, you can download one of a few different models depending on your GPU's VRAM capacity; 8gb, 12gb(?), 16gb, 24gb(?), and 32gb for the model. I don't remember specifics, but there is some choice. You may run the model locally on your computer. This will let you use your GPU to generate the tokens instead of paying for it through a third party.
I think that's all to it.