Cost doesnt add up?
# support
e
v
Hi @Ethan Tan, I'm an AI assistant for Vapi. I'll analyze available sources and get back to you shortly.
----------------------------- **While I generate an answer, here are some sources that might be helpful**: * [Estimating Costs]() * [Billing Examples]() * [Pricing Overview]() * [Custom LLM]() -----------------------------
### Answer To address your question about the cost discrepancy, it's essential to consider that the total cost in Vapi may include additional components not immediately visible in your breakdown. Vapi charges for various elements such as the platform fee, transcription (STT), model inference (LLM), and text-to-speech (TTS), along with any telephony costs if applicable. Each of these elements could be contributing to the total cost you're seeing. The charges for these services can also vary based on the providers you're using, such as Openrouter for the LLM and any specific settings or configurations that might increase usage or complexity, thus affecting the cost. Unfortunately, without specific details on the call minutes or the exact configuration of your setup from the provided documents, I cannot give an exact breakdown or reason for the discrepancy you're observing. For a detailed understanding of how costs are estimated and broken down, you might want to refer to the Vapi documentation on estimating costs, which could provide some insights into how these calculations are made and why there might be differences in the expected vs. actual costs [(1)]() .
s
Can you send me the call_id?
e
Hi @Sahil yes its c581569a-3c17-4865-bb6c-e06a5a16172c Thank you
j
It shows the total cost, and the cost per minute. Your call would have been closer to 2 minutes total
s
^ @Ethan Tan he is correct your call duration was 1.73 min. So, the cost is accurate.
e
Thanks @Sahil and @Jotham - I get that, however why does the breakdown not reflect the actual price? STT: 0.02 LLM: 0.00 TTS: 0.05 Vapi: 0.09 Transport: 0.00 Total: 0.24
j
Looks like the LLM pricing is off, are you using your own API keys for the LLM? Which model are you using
s
Calculate based on 1.73 minutes and the price is accurate.
e
@Sahil it's the LLM price that's not showing - the other parts show both per minute and total @Jotham that was with Llama-3 70b on Openrouter
s
Can you tell me what the cost was for LLM model for this call from the openrouter side?
e
@Sahil it was about $0.06
Yes I was using Openrouter with my API keys - is there an option to use it with Vapi's keys?
@Sahil @Jotham Following up
s
Let's debug it together, sent you an meeting invite.