The AI model currently used is OpenAI gpt-4-turbo (gpt-4-0125-preview). Please note that GPT-4 models are more expensive than previous GPT models.
Currently and very approximately considering the complexity of the cost calculation, we estimate, for example, that getting results from Kai for scenarios with one or two initial short exchanges, then getting a scenario narrative could cost around 200 RTAS credits (approximately 12,5 times more than with GPT 3.5).
This means that with the Light Access you should be able to get 12 narratives. With the Trial Access you should get approximately 5 narratives.
If you generate one image this would cost around 40 credits per image.
Each exchange (one question and one answer) with any AI assistants could cost between 40 and 100 credits.
Please note that these are not exact costs and are only given as an order of magnitude.