AI Voice Agent Cost Per Minute: $0.09 to $0.21 All-In

Short answer: an AI phone agent on US lines costs about $0.09 to $0.15 per minute on a budget stack and up to about $0.21 per minute with a frontier LLM and premium voices. At 1,000 minutes a month that is roughly $90 to $150; at 10,000 minutes, roughly $900 to $1,500. The platform fee you see on the pricing page is only one of five layers.

Every voice agent pays for the same five things: the platform that orchestrates the call, speech-to-text (STT), the LLM that decides what to say, text-to-speech (TTS), and telephony (the phone line). Some vendors bundle layers and some itemize them. Below, each stack is built from the vendor’s own pricing page, checked 28 October 2026 (Twilio checked 29 October 2026). All prices are USD and exclude tax.

The five layers, and who charges for what

Sources: Vapi, Retell, ElevenLabs Agents, Bland.

Telephony baseline used below: Twilio’s US rate to receive calls on a local number is $0.0085 per minute, to make calls $0.0140 per minute, plus $1.15 a month per local number (Twilio US). Canada is identical (Twilio CA). Vapi’s page lists Twilio inbound at $0.008 per minute; we use Twilio’s own published $0.0085.

Per-minute stack, vendor by vendor

Vapi

Vapi charges $0.05 per minute for hosting on the no-commitment plan and passes models through at cost. Its own calculator at 1,000 minutes shows:

Without telephony, that matches Vapi’s calculator estimate of $82 to $129 a month at 1,000 minutes. If you use Vapi’s own number (one included on pay-as-you-go, five on Core), Vapi lists telephony as free, which drops about $0.0085 a minute.

Plan notes: pay-as-you-go allows 4 concurrent calls and 14-day data retention. Core is $29 a month for 10 concurrent calls and 5 numbers. Pro has a $999 monthly minimum. HIPAA is a $2,000 a month add-on, and extra concurrency is $10 per line per month.

Retell AI

Retell bills per second and lists pay-as-you-go agents at $0.07 to $0.31 per minute. You pick each layer:

Retell does not charge for SIP or custom telephony, so bringing your own Twilio trunk can replace its $0.015 line with your carrier’s rate. Phone numbers are $2 a month. You get 20 concurrent calls free, then $8 a month each. Add-ons stack up: knowledge base, denoising and guardrails are +$0.005 per minute each, PII redaction +$0.01. Retell also notes that billing continues during silence.

ElevenLabs Agents

ElevenLabs bills agents by call minute: $0.08 per minute beyond your plan’s included minutes, which are themselves priced at $0.08 (Creator: $22 for 275 minutes). The LLM is billed separately, and telephony goes straight to your carrier.

ElevenLabs does not show a single per-minute LLM rate, so we use the same range as the other stacks: $0.008 (budget, matching Retell’s GPT 5 mini) to $0.045 (frontier, matching the top of Vapi’s OpenAI range). Treat this as an assumption.

Burst calls above your concurrency limit cost $0.16 per minute. For the full plan table and a podcast example, see our ElevenLabs pricing per minute guide.

Bland AI

Bland bundles the LLM, STT and TTS into one rate. Start is $0.14 per minute with no platform fee, 10 concurrent calls and a cap of 100 calls a day. Build is $0.12 per minute plus $299 a month, with 50 concurrent calls and 2,000 calls a day. Transfer minutes are $0.05 (Start) or $0.04 (Build). Telephony is billed separately.

The trade-off is control: you do not pick the model or voice per layer, and on Start the 100-calls-a-day cap limits you to about 3,000 calls a month.

Monthly totals at 1,000 and 10,000 minutes

US inbound calls on Twilio, per-minute rates from the tables above, platform fees included where required. Phone number rental ($1.15 to $2 a month) is left out.

How the 10,000-minute numbers work:

• Vapi: 10,000 × $0.0903 = $903. At that volume most teams need more than 4 concurrent calls, so we add Core at $29.
• ElevenLabs: the ElevenLabs share is 10,000 × $0.08 = $800 whichever plan you start from (Pro: $99 + 8,762 extra minutes × $0.08 = $799.96). Add $80 LLM and $85 telephony = $965.
• Bland Build vs Start: Build saves $0.02 a minute but costs $299 a month, so it breaks even at $299 ÷ $0.02 = 14,950 minutes. Below that, Start is cheaper if its daily call cap fits.

The ranking flips with your choices. Retell’s budget stack beats ElevenLabs Agents, but Retell with premium ElevenLabs voices costs more than ElevenLabs’ own agent. Pick the LLM and voice first, then compare platforms.

Canada and Australia

Canada: Twilio’s Canadian rates match the US ($0.0085 to receive, $0.0140 to call, $1.15 per local number), except calls to the Yukon at $0.1525 per minute. Stack totals above carry over unchanged.

Australia: telephony is pricier, especially for outbound. Twilio charges $0.0100 per minute to receive on local or mobile numbers, $0.0252 per minute to call Australian landlines and $0.0750 per minute to call mobiles, plus $3.00 a month per local number (Twilio AU). An outbound sales agent dialing mobiles adds about $0.066 a minute versus US inbound, which can raise a budget stack from about $0.09 to about $0.16. Retell’s $0.015 telephony rate is shown for US Twilio; for Australia, plan on your own SIP trunk and Twilio’s AU rates.

In local currency (ExchangeRate-API, 29 October 2026: 1 USD = 1.40 CAD = 1.42 AUD): a $93 Retell budget month is about CA$130 or A$132; a $965 ElevenLabs budget month at 10,000 minutes is about CA$1,353 or A$1,370.

Tax: these vendors quote prices before tax. Australia’s 10% GST applies to imported digital services sold to Australian consumers, with different treatment for GST-registered businesses (ATO). Canada’s GST/HST digital-economy rules have applied to foreign vendors since July 1, 2021 (CRA). If you bill through a US entity, see our guides for Canadians and Australian founders. This is general information, not tax advice.

Be careful with vendor comparisons

Most “Vapi vs Retell vs Bland” pages are written by one of the vendors. Two examples:

• Bland’s pricing FAQ says production Vapi stacks run $0.13 to $0.30 per minute and Retell $0.11 to $0.25. Vapi’s own calculator shows a budget stack at about $0.08 before telephony, and Retell’s own rate card allows $0.093 including telephony. Bland’s numbers are not impossible (premium models, add-ons, outbound calls), but they describe the expensive end.
• Retell publishes its own head-to-head of Retell, Bland, Vapi and ElevenLabs. Read it as marketing, not a benchmark.

The fix is to rebuild every quote from the five layers yourself, as above, using the same LLM, voice and phone route on each.

Which platform fits which team

Two other options appear on pricing pages but are not compared here: Deepgram’s Voice Agent API at $0.075 per minute (Deepgram) and Twilio’s ConversationRelay at $0.07 per minute. Check what each includes before comparing.

How to cut your per-minute cost

• Choose the LLM deliberately. On Retell, GPT 5 mini is $0.008 a minute and GPT 6.1 Sol is $0.064. That one choice is worth $0.056 a minute, or $560 at 10,000 minutes.
• Use cheaper voices for routine calls. Retell’s own voices are $0.015 versus $0.08 to $0.10 for ElevenLabs v3/v4.
• Bring your own telephony where the platform charges a markup or nothing for SIP.
• Keep calls short. Retell bills during silence; tighten prompts and hang up cleanly.
• Watch add-ons. Knowledge base, denoising and guardrails at +$0.005 each add $0.015 a minute together.

Bottom line

On list prices, Vapi, Retell and ElevenLabs Agents all land around $0.09 to $0.10 per minute for a budget US stack, and roughly $0.13 to $0.21 with frontier models and premium voices. Bland’s bundled $0.14 is simpler but dearer at low volume. Your LLM, your voice and your phone route matter more than the platform. For more sourced, dated cost breakdowns, see the AI tools hub, or compare automation tools in our n8n vs Make vs Zapier pricing guide. If text chat is enough, a WhatsApp chatbot on n8n costs cents per conversation, and our n8n AI agent cost per run guide breaks down the token math.

The LLM line is the easiest one to tune: compare 12 current models on your own token counts with the free LLM API Cost Calculator.