Currently, all AI models (chat, & image) count tokens at the same rate. e.g. 1 token for say gpt-4 is the same as 1 token for gpt-3.5-turbo. Ideally, we would be able to set a custom token ratio per model so that we can charge more for some models or even discount the cost of some models.
Currently, all AI models (chat, & image) count tokens at the same rate. e.g. 1 token for say
gpt-4is the same as 1 token forgpt-3.5-turbo. Ideally, we would be able to set a custom token ratio per model so that we can charge more for some models or even discount the cost of some models.