High Input Tokens Usage (Using own Claude API Keys)

I decided to use my own keys as I have exhausted the plan limits. And I noticed that I just noticed that for each request, the request token size is huge.

Am I using it wrong? Is it a bug? What can I do to prevent this? or is this how this is supposed to work.

I just burned through $$ pretty quick after a long coding session.

Any inputs on this would be of great help!

I think it’s because of the many tokens that AI has to read, so it costs a lot.

Most of the time I am just working on one or at max two files. How could that end up being 45K+ tokens consistently for every request? The files are max 200 or 300 lines of code. Def nothing close to 45,000 tokens.

i dont know but i think you should create a new Cursor account and buy premium lol

I have premium. Used 500 fast requests > Then bought another 500 which I finished again. This is where I decided to plug in my own keys. There definitely something I am doing wrong or I am missing something. Let’s see if people who have actually used their own API keys have any real answer to this riddle :slight_smile: