I would like to know whether in e.g. Gemini-2.5 pro we pay for input and output tokens, and the output tokens are only output tokens or the output tokens are output tokens plus thinking tokens?
Yes, you pay for Thinking Tokens (in addition to the Output tokens).
You can set thinkingBudget: number in the config to control how many are used. The minimum for Gemini 2.5 Pro is 128, the maximum is 32768. This all counts towards the max output token limit.
I would like to have a certain answer because my tests show that we do not pay for tokens for thinking.
Hi @Tomasz_Swiercz
Check this out Gemini Developer API Pricing | Gemini API | Google AI for Developers.
Output price is including thinking tokens.
Thank you