Challenges with WebRTC Stream APIs in Terms of Credit Consumption and Affordability

I would like to inquire about how the usage of WebRTC stream APIs affects the consumption of credits. Does it calculate based on the duration of the WebRTC session or the time it takes to generate the speech? Either way, according to my calculations, a basic chatbot session would cost approximately $0.6 to $2.5. This ongoing cost is not feasible for many scenarios. For instance, we are developing a mental health chatbot solution and while we love and appreciate the D-ID stream solution, we are unable to afford the ongoing cost for our clients. It poses a challenge for us.
Thanks