A voice agent is three bills in one: speech-to-text + LLM + text-to-speech, all running per minute of conversation. The LLM and premium voices dominate. Enter your volume to see the real monthly cost.
—
per month—per call
—per minute
—total minutes/mo
Where the money goes (per minute)
| Component | $ / min | % of total |
|---|
⚠️ Reference estimates (July 2026). STT ~$0.006/min (Whisper-class), TTS $0.015–0.30/min (premium neural voices cost more), LLM depends on tokens per turn — type your real numbers. Telephony/SIP minutes are extra if you connect a phone number. · Report outdated price →
How voice agent cost adds up
Per minute you pay for STT (transcribe the caller), the LLM (decide the reply — token-based, so longer context and answers cost more), and TTS (speak it back). The LLM and a premium voice usually dominate; STT is the cheapest leg. Estimate the LLM leg precisely with the AI agent cost calculator or LLM token cost, and compare TTS providers in our ElevenLabs and Deepgram guides. Building a text chatbot instead? Use the chatbot cost calculator.