Reported Token Discrepancies

@visarg can take a look at this

Thank you, and one more thing for reference I realized; my Gemini API is linked to my Google Cloud Console, which is on a trial with free credits. If that helps in any way

So I tested some more:

It was the same command that I ran. Now OpenAI's token count seems much more accurate, and the costs are accurate too.

I know Gemini's tokenizer may be different, but I don't think it should be 3-4x higher, and I don't see any way to check the logs on the Gemini console.

Gemini's tokens are characters - so you might be seeing that

We'll get the costs fixes soon!

Ah right, that makes sense, thank you!