Why are my calls suddenly much higher?

Support Help
1

I have been testing my app but suddenly the price has gome up event though exact same LLM and voice. at top it says 13.6 cents buit for example this less than 4 minute call was 76 cents call_e9e97ecd302c65a5702c75debde

here is another 3 minute 63 cents call_c3382e18b2c3f2fbf3144179e51 Any help appreaciated

2

Hey @ken2

Thank you for the details. I’ve forwarded this to our team for review.

We’ll get back to you as soon as we have an update.

Best regards

3

Hello @ken2

The “13.6¢/min” shown at the top of the dashboard is the base rate (voice engine + TTS + LLM + telephony). Your two calls are hitting an additional LLM token surcharge, which is why they look ~5× more expensive than the headline rate.

Here’s the exact breakdown from our billing records:

call_e9e97ecd… (229s, GPT-5.1) — $0.7391 total

  • Voice engine: $0.210
  • Platform TTS: $0.057
  • GPT-5.1: $0.153
  • US Twilio telephony: $0.057
  • LLM token surcharge: $0.262

call_c3382e18… (173s, GPT-4.1) — $0.6358 total

  • Voice engine: $0.159
  • Platform TTS: $0.043
  • GPT-4.1: $0.130
  • US Twilio telephony: $0.043
  • LLM token surcharge: $0.261

The surcharge is triggered when the average input tokens per LLM request on a call exceeds 3,500. Above that threshold, billed LLM duration is scaled by tokens / 3,500, so the more context you send per turn, the higher the surcharge.

Your two calls averaged:

  • Call 1: 9,505 input tokens / request (~2.7× the threshold)
  • Call 2: 10,531 input tokens / request (~3× the threshold)

Thank You

4

can you tell me what I changed that made it so much more expensive? Is it the use of tools?

5

Hi @ken2

The platform has a 3,500 input token threshold per LLM request. Once your average exceeds that on a call, a token surcharge kicks in and scales your billed LLM duration. it’s not which model you used, it’s how many tokens you’re sending per turn that made it expensive.

Thank You

6

Thanks much.
So is there any way to get notice if this happens ?

Kendall Stauffer

7

Hi @ken2

There’s no direct alert specifically for the 3,500-token threshold, but you can set a Total Call Cost alert via the Alerting feature to get notified (email or webhook) when costs exceed a threshold which would catch the scaling surcharge’s impact. Available metrics don’t include prompt token length directly.

Thank You