The latest version of gpt-3.5-turbo is very inexpensive to run, gpt-3.5-turbo-1106 input $0.0010 / 1K tokens and output $0.0020 / 1K tokens. If you can purchase $1 of credits then you can run a lot of API calls before you reach your limit. There are opensource alternatives as well, but you will need to know or learn how to run them locally,

That is not true. With a cost of one coffee you can do tens or evem hundreds of queries. The API based access is incredibly cheap. If you cannot afford that, but still have a computer, then you should do two things: 1) take a deep look at your life choices and 2) learn to use llama models locally.

I used two million tokens in the last two days. That's two Euros and a lot of data

Yeah money i dont have cuz im really broke