This is a very silly question, but it’s my first time playing around with this technology.
Do the input/output tokens carry over between requests?
So assuming 1mil token are $0.10. If I have 10 separate 100,000 token inputs, do I pay $1 or $0.10?
2 Likes
Welcome to the forum. The input tokens (and the output tokens, for that matter) are cumulative. So, 10 100k requests would theoretically be equivalent to a 1M request.
For many models, however, there is a 128k threshold beyond which the tokens get more expensive, so for those models the price of 10 100k requests is in fact less than the one large request. You should refer to the pricing page at Gemini Developer API Pricing | Gemini API | Google AI for Developers for the model-specific details.
Hope that helps
1 Like