Voice agent cost calculator
Enter your call minutes and pick the parts. See what an AI phone agent costs per minute and per month on Vapi, Retell, LiveKit Cloud, ElevenLabs, Bland, or your own servers.
Model token assumptions
Every turn resends the instructions, tools, and the call so far, so input tokens grow through a call. These defaults are our assumption for a short booking call, not a measurement.
Monthly bill on Vapi
| Line | Per minute | Per month |
|---|---|---|
| Vapi platform | $0.0500 | $150.00 |
| Speech to text: Deepgram Nova-3 streaming | $0.0077 | $23.10 |
| Model: GPT-5.4 mini | $0.0078 | $23.42 |
| Voice: ElevenLabs Flash | $0.0125 | $37.44 |
| Phone line: Twilio, inbound to a US local number | $0.0085 | $25.50 |
| Plan, servers, and phone numbers | $1.15 | |
| Total | $0.0869 | $260.61 |
List prices, pay as you go, US calls. Voice prices checked 2026-10-13, model prices checked 2026-10-21. The total per minute includes the monthly fees spread over your minutes. Taxes, volume discounts, extra concurrency, and build or upkeep work are not included.
The default scenario, worked out
Inbound US calls on Twilio to a US local number, Deepgram Nova-3 streaming for speech to text, GPT-5.4 mini as the model, and ElevenLabs Flash as the voice, wherever the platform lets you choose. Retell uses its own listed rates for GPT 5.4 mini and its platform voices. One phone number. Plan fees as listed in the table.
| Platform | Usage per minute | 1,000 min / month | 3,000 min / month | 10,000 min / month |
|---|---|---|---|---|
| Vapi | $0.086 | $88 | $261 | $866 |
| Retell AI | $0.103 | $104 | $309 | $1,026 |
| LiveKit Cloud | $0.036 | $88 | $161 | $466 |
| ElevenLabs Agents | $0.063 | $116 | $290 | $964 |
| Bland AI | $0.149 | $150 | $447 | $1,486 |
| Custom, self-hosted | $0.036 | $138 | $211 | $466 |
Usage per minute is before monthly fees. LiveKit Cloud is on Ship ($50 a month, 5,000 minutes included). ElevenLabs Agents is on Pro ($99 a month, 1,238 minutes included). Self-hosted adds a $100 a month server placeholder. Model tokens assume 6 turns a minute, 2,500 input and 60 output tokens a turn, and 50% of input from the prompt cache. Voice prices checked 2026-10-13. Model prices checked 2026-10-21.
How the number is built
Each call minute pays for up to five things: the platform, speech to text, the model, the voice, and the phone line. Platforms bundle some of these. Bland includes all three AI parts. ElevenLabs includes speech and voice but bills the model on top. Retell bills the model and voice at its own per-minute rates. Vapi and LiveKit pass provider prices through.
The voice is priced by characters spoken. Cartesia says one minute of audio takes 750 to 800 characters, so we use 780 and scale it by how much of the call the agent is talking.
What it leaves out
The work of building, testing, and changing the agent. Extra concurrency, recordings, HIPAA add-ons, taxes, and volume discounts. The model token numbers are our assumption for a short booking call, not a measurement, so open the token settings and change them for your calls.
For what the parts of a voice agent do, read what a realtime voice agent consists of.
Questions clients ask
Short answers, no sales language. If yours isn't here, ask an engineer directly.
Ask us anythingHow much does an AI phone agent cost per month?
On list prices, a US inbound line with a small model costs about 4 to 15 cents a minute, depending on the platform. At 3,000 minutes a month that is roughly $161 to $447, before the cost of building and testing the agent.
Why does the platform fee matter so much?
Because the parts underneath are cheap. Speech to text, a small model, a voice, and a phone line add up to about 3 to 4 cents a minute at list price. A platform fee of 5 cents a minute is more than all of them together.
Is a realtime speech-to-speech model cheaper?
Usually not at list price. Retell lists GPT Realtime at $0.345 a minute and GPT Realtime mini at $0.07 a minute, which replace the model and the voice. Compare those with the model and voice lines in the calculator.
What is not in the number?
Building, testing, and changing the agent, extra concurrency, call recording storage, taxes, and volume discounts. Self-hosting has no platform fee, but someone has to run the servers. The $100 a month for servers is a placeholder.
Planning a phone agent?
We build AI agents with approval steps and a handoff to a person. An engineer replies, not a sales team.