See AloAi voice agent in action: test our AI voice agent and see how it automates call handling at scale.

Talk to agent

Pay by the minute.

Launch by the afternoon.

The fastest way to put a production-grade AI voice agent on a phone number. Every component priced transparently, billed by the second, with no minimum commitment and no per-seat fees. Bring your own LLM, voice, and telephony — or use ours.

Book a 15-minute call
on
by 300+ verified sales teams.
10,000+ teams
Setup in 10 minutes
Cardone Venturescarvioclosers.ioClub NewportCoding DOJOCommunity laborAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisorist
Cardone Venturescarvioclosers.ioClub NewportCoding DOJOCommunity laborAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisoristAdvisorist
AI Voice Agent Pricing Per Minute — From $0.07/min, No Lock-in
The short answer Featured snippet · FAQ
Q.

How much does an AI voice agent cost per minute?

AI voice agent pricing per minute typically ranges from $0.07 to $0.19 per minute, fully loaded with LLM inference, text-to-speech, telephony, and voice infrastructure. CapacityAgents starts at $0.07/min for a Gemini 2.5 Flash Lite + platform-voice + SIP configuration and scales to about $0.19/min for a Claude 4.5 Sonnet + ElevenLabs + Twilio configuration. You only pay for the seconds the agent is on a call — no idle time, no per-seat fees, no minimums.

01
Billing unit
per second, displayed as per minute
02
What's included
LLM, TTS, ASR, voice infra, telephony, transcripts, recordings, dashboards
03
Floor / typical / ceiling
$0.07/min · $0.10–$0.13/min · ~$0.19/min
04
Concurrency
20 simultaneous calls included, more on demand
05
Minimum commitment
none — pay-as-you-go
AI Voice Agent Pricing Per Minute — From $0.07/min, No Lock-in
Head-to-head

Per-minute vs. outcome-based AI voice agent pricing.

Per-minute
Outcome-based
You pay for
Every second of audio
Verified business result
Time to launch
Same day
2–4 week pilot
Cost predictability
Tied to call volume
Tied to conversion rate
Best for
Pilots, experimentation, variable use cases
Defined funnels at scale
Minimum commitment
None (pay-as-you-go)
Monthly outcome floor
Vendor incentive
Uptime and quality
Hit your KPIs
Easy to switch stacks
Yes — anytime
Locked for contract term
Easy to forecast
When call volume is stable
When conversion rate is stable
Per-minute pricing is the right model when call volume is variable, use cases are still being defined, or the team needs to iterate weekly on prompts and voices. Outcome-based pricing is the right model when the funnel is mature and the buyer wants vendor accountability for results.
Already know your conversion rate? See outcome-based pricing →
AI Voice Agent Pricing Per Minute — From $0.07/min, No Lock-in
Speed & flexibility

When you're moving fast, per-minute is the right shape.

Outcome-based pricing is great when you already know your funnel cold. Per-minute is great when you're still finding it. You can spin up an agent today, tune it tomorrow, and replace half the stack next week — without renegotiating a contract.

Launch the same day

No pilot phase, no SOW, no procurement. Sign up, point a number, ship a prompt. Most teams have an agent answering live calls within 4 hours.

Total component transparency

Every minute is broken down into LLM cost, TTS cost, telephony cost, and voice infra. You can see exactly which knob to turn to drop your bill.

Swap anything, anytime

Change LLMs mid-day. A/B test voices. Switch from Twilio to Telnyx for a region. Bring your own SIP trunk. Your cost-per-minute updates instantly.

AI Voice Agent Pricing Per Minute — From $0.07/min, No Lock-in
Deliverability

Cheap per-minute pricing is worthless if the agent can't hold a conversation.

A lot of "$0.05/min" claims fall apart the moment a real caller interrupts the agent. Per-minute pricing only matters if every minute is a minute the caller stays on the line. Here's how we make sure they do.

Sub-400ms response latency

End-to-end ASR → LLM → TTS measured continuously across regions. Callers don't get the AI tell-tale pause.

Real-time interruption handling

Barge-in detection, half-duplex audio, and recovery from cross-talk. The agent stops talking the instant the caller does.

Carrier-grade telephony

Native Twilio, Telnyx, Bandwidth, and BYO-SIP. STIR/SHAKEN attestation. Local presence dialing in 40+ countries.

Built-in fallback system

Automatic failover across LLM providers, voice providers, and carriers. A model outage doesn't drop your calls.

Production observability

Live transcripts, audio recordings, latency traces, sentiment scores, and call success metrics in a single dashboard.

Engineering-grade SDK

REST API, webhooks, function calling, custom tool definitions, and CRM connectors. Treat the agent like any other microservice.

<400ms
median agent response latency
99.95%
platform uptime SLA, contractually backed
20
concurrent calls included on every account
40+
countries with local-presence telephony
15+
LLMs, 6 voice providers, 8 telephony carriers ready to wire up
AI Voice Agent Pricing Per Minute — From $0.07/min, No Lock-in
What's in a minute

Every cent on every minute, visible.

A per-minute price is only honest if you can see the components. Below is the exact stack pricing — the same numbers our cost calculator uses.

Always included
Voice infrastructure
ASR, audio routing, session management, recording, transcript
$0.055/min
LLM options pick one
GPT 4.1$0.045
GPT 5.4$0.080
GPT 5.2$0.056
GPT 5.1 / GPT 5$0.040
GPT 5 mini$0.012
GPT 5 nano$0.003
GPT 4.1 Mini$0.016
GPT 4.1 Nano$0.004
Claude 4.6 Sonnet$0.080
Claude 4.5 Sonnet$0.080
Claude 4.5 Haiku$0.025
Gemini 3.0 Flash$0.027
Gemini 2.5 Flash$0.035
Gemini 2.5 Flash Lite$0.006
Custom LLM (BYO)$0.000
Text-to-speech pick one
Platform Voices$0.015
Minimax$0.015
Fish$0.015
Cartesia$0.015
OpenAI$0.015
ElevenLabs$0.040
Telephony pick one
Custom (BYO SIP)$0.000
Twilio$0.015
Telnyx (US)$0.015
Worked example · recommended default stack
$0.055 Voice infra + $0.045 GPT 4.1 + $0.015 Platform Voices + $0.015 Twilio US
Total $0.130/min
Build your own stack → Use the cost calculator

Estimate your cost

Pick your LLM, voice, and telephony. See the real per-minute rate for your setup.

Enter your expected monthly call volume in minutes.

Cost Per Minute
$0.000
Estimated for the per-minute plan
Total per month $0.00

For the Agents by Minute plan

Block 06 · Billing Transparency — Preview
Billing transparency

How the meter runs.

No rounding tricks, no surge surprises. Three things to know.

Step 01
Connect

The clock starts when audio flows

Billing begins when the call connects and audio starts. Ringing, busy signals, voicemail detection, and unanswered calls cost $0.

Step 02
Per-second

Billed by the second, displayed by the minute

A 47-second call is billed for 47 seconds — not rounded up to a minute. Most platforms round; we don't.

Step 03
Itemized

One invoice, four line items

Your monthly invoice shows total minutes, plus a breakdown by LLM, TTS, telephony, and voice infra. Reconcile against your usage logs in one click.

No surge pricing. Concurrency above 20 simultaneous calls bills at the same per-minute rate. No overage penalties.
AI Voice Agent Pricing Per Minute — From $0.07/min, No Lock-in
Who it's for

Is per-minute pricing right for you?

Per-minute is a fit if you:

  • Are launching your first AI voice agent
  • Run multiple use cases with different success criteria
  • Iterate on prompts, voices, or flows weekly
  • Have variable or unpredictable monthly call volume
  • Want to bring your own LLM, voice, or telephony
  • Need to avoid procurement and contract overhead

Consider outcome-based if you:

  • Run 10,000+ qualifying or support calls per month
  • Already track cost-per-acquisition or cost-per-resolution
  • Need fixed unit economics for the board
  • Want the vendor on the hook for performance

Questions teams ask before switching

How much does an AI voice agent cost per minute?

Production-grade AI voice agents run $0.07–$0.19 per minute fully loaded, depending on LLM, voice, and telephony choices. CapacityAgents' recommended default stack — GPT 4.1, Platform Voices, Twilio US — is $0.13/min. The lowest viable production stack is $0.07/min using Gemini 2.5 Flash Lite, Platform Voices, and BYO SIP.

Is there a free trial?

Yes. New accounts get free credits to test live calls. No credit card required. Concurrency, voices, LLMs, and analytics are unrestricted during the trial.

Are there per-seat or setup fees?

No. Per-minute pricing is purely usage-based. No seats, no setup fees, no SSO surcharge, no support tier upcharge. Twenty concurrent calls are included on every account.

How do you bill — per second or per minute?

Per second, displayed as per minute. A 47-second call bills for 47 seconds. Most vendors round up to the nearest 6 or 60 seconds; we don't.

What's the cheapest production-ready configuration?

$0.07/min using Gemini 2.5 Flash Lite ($0.006), Platform Voices ($0.015), BYO SIP telephony ($0.00), and our voice infrastructure ($0.055). It's the cheapest stack we'd actually deploy in production.

Can I bring my own LLM, voice, or telephony?

Yes to all three. BYO LLM is $0.00/min on the LLM line. BYO SIP telephony is $0.00/min on the telephony line. You're billed only for the components you don't supply, plus the $0.055/min voice infra.

What happens if a call goes over 20 concurrent?

Calls above 20 concurrent bill at the same per-minute rate — no surge pricing. If you need a hard reservation for a campaign, we can pre-allocate dedicated concurrency on request.

Are recordings, transcripts, and analytics extra?

No. They're included in the $0.055/min voice infrastructure line. Storage retention is 90 days by default; longer retention is available.

What's the latency like?

Median end-to-end response latency is under 400ms across our default stack and regions. Latency dashboards are available in real time.

How do I move from per-minute to outcome-based later?

A per-minute account is the natural starting point for an outcome-based contract. The conversion rate, handle time, and cost data from your per-minute usage become the inputs to your outcome price.

Is there a minimum monthly spend?

No minimum. Pay only for what you use.
AI Voice Agent Pricing Per Minute — From $0.07/min, No Lock-in

Spin up an agent in 15 minutes.

Walk us through one use case and we'll wire up a live agent on a real number — on the call. Bring a phone; you'll hear it talk before you hang up.

Book a 15-minute call → Start free in the dashboard

Talk to a voice engineer, not a salesperson. No credit card required to start.