Cursor Consuming 2 Fast requests for 1 prompt

I was looking at one point at my fast requests but I noticed that for a single prompt, the cost is 2 fast requests. How is this possible. It clearly states everywhere that one message=1 fast request. So why is it using 2?w

Turning on"thinking" make some models go x2 requests

Yeah i noticed that, but i did not turn on thinking. Also, when I’ve tried it with Auto-select, same exact result. When i was on a free account, it didnt do that, but as soon as i got the pro, starte using 2 requests per message.