How much does an AI voice agent cost per minute?
On the six AI voice agent platforms we checked on October 17, 2026, a standard setup costs about $0.08 to $0.14 per minute in usage fees, including the phone line. Premium voices and larger language models push that past $0.23 per minute. Building and maintaining the agent is a separate cost.
AI voice agent pricing is hard to compare because each vendor puts different parts into "a minute". Some quote one bundled rate. Others charge a platform fee and pass through speech, language model and phone costs. So we read each pricing page, wrote down every published rate, and priced the same month of calls on each platform. The method and every source are at the bottom of the page.
What this page adds: one table of list prices read on the same day, and one worked example (1,000 minutes of inbound calls) costed the same way on every platform, so the totals can be compared.
What are the published AI voice agent prices, platform by platform?
HighLevel Voice AI charges a $0.045 voice engine rate plus voice and model costs, or a flat $97 a month. Retell starts at a $0.055 infrastructure rate, Vapi at a $0.05 hosting fee, Bland at $0.14 a minute, and ElevenLabs sells bundles of minutes from $22 a month. Synthflow now quotes enterprise contracts only.
| Platform | How it charges | Published rates, read October 17, 2026 |
|---|---|---|
| HighLevel Voice AI (pay per use) | Voice engine + text-to-speech + model tokens; phone charges separate | Voice engine $0.045/min (from May 20, 2026). Voices: OpenAI or Cartesia $0.015/min, ElevenLabs V2.5 $0.035/min, ElevenLabs V3 $0.170/min. Tokens per model, e.g. GPT-5 mini $0.25 in / $2.00 out per million |
| HighLevel Voice AI (speech-to-speech) | One bundled rate per minute; phone charges separate | Gemini 3.1 Flash Live preview $0.10/min; OpenAI gpt-realtime-2/2.1 $0.20/min |
| HighLevel AI Employee plans | Per location per month; phone charges separate | Growth $50/month with 100 Voice AI minutes, overage at pay-per-use rates. Unlimited $97/month, unlimited Voice AI subject to fair use |
| Retell AI | Pay as you go, itemised | Voice infrastructure $0.055/min. Voices $0.015/min (Retell, Cartesia, OpenAI and others) to $0.10/min (ElevenLabs v3 and v4). Models from $0.0016/min (GPT 5 nano) to $0.32/min (GPT 6 Astra, standard tier). Telephony $0.015/min, none with your own SIP trunk. Knowledge base +$0.005/min. Overall range stated as $0.07 to $0.31/min |
| Vapi | Hosting fee + providers at cost | Hosting $0.05/min. Calculator ranges: Deepgram transcription $0.0095 to $0.0099/min, OpenAI models $0.0077 to $0.0452/min, ElevenLabs voices $0.0146 to $0.0238/min. Twilio transport $0.008/min inbound, $0.014/min outbound. Optional packages: Core $29/month; Pro 10% of hosting with a $999/month minimum |
| Bland | Per connected minute by plan | Start: no platform fee, $0.14/min, transfers $0.05/min. Build: $299/month, $0.12/min, transfers $0.04/min. Enterprise custom |
| ElevenLabs Agents | Monthly plans with included call minutes; model billed separately; your phone provider bills you directly | Creator $22/month with 275 minutes; Pro $99 with 1,238; Scale $299 with 3,738; Business $990 with 12,375. Extra minutes $0.08 ($0.16 burst above your concurrency limit) |
| Synthflow | Enterprise contracts | "Enterprise contracts start at $30,000 annually", scoped by volume, concurrency and integrations |
Prices are list prices for US calls, in US dollars, without promotions. ElevenLabs was running first-month discounts on the day we checked; the table shows the regular monthly prices.
What does a 1,000-minute month cost on each AI voice agent platform?
For 1,000 minutes of inbound US calls, about 250 calls of four minutes, the standard setups we priced land between $76 and $140 a month in usage fees. HighLevel pay-per-use was lowest at about $76, Bland's Start plan highest at $140. Premium voice setups cost about $230.
A 1,000-minute month on each platform
How we priced it, so you can redo it with your own numbers:
- Volume: 1,000 minutes of inbound calls on a US local number. Illustrative, not a client's data.
- Standard setup: the cheapest standard voice each platform lists, a small language model, and the platform's own phone line or Twilio.
- Language model where it is billed by tokens (HighLevel pay-per-use, ElevenLabs): we added $0.008 per minute, which is Retell's listed per-minute price for GPT 5 mini. Your real token cost depends on prompt length and how much the caller talks.
- Phone line: HighLevel inbound $0.00765/min; Twilio inbound $0.0085/min for ElevenLabs, $0.008/min as Vapi lists it; Retell's own $0.015/min. Bland's per-minute rate was taken as listed.
| Platform and setup | Calculation | 1,000 minutes |
|---|---|---|
| HighLevel Voice AI, pay per use, OpenAI voice | (0.045 + 0.015 + 0.008 model) x 1,000 + phone $7.65 | $75.65 |
| Vapi, cheapest calculator options | (0.05 + 0.0095 + 0.0077 + 0.0146 + 0.008) x 1,000 | $89.80 |
| Retell, Retell voice, GPT 5 mini | (0.055 + 0.015 + 0.008 + 0.015) x 1,000 | $93.00 |
| ElevenLabs Agents, Creator plan | $22 + 725 extra minutes x $0.08 + model $8 + Twilio $8.50 | $96.50 |
| HighLevel AI Employee Unlimited | $97 flat + phone $7.65 | $104.65 |
| HighLevel speech-to-speech, Gemini Live | 0.10 x 1,000 + phone $7.65 | $107.65 |
| Vapi, top of calculator ranges | (0.05 + 0.0099 + 0.0452 + 0.0238 + 0.008) x 1,000 | $136.90 |
| Bland, Start plan | 0.14 x 1,000 | $140.00 |
| HighLevel pay per use, ElevenLabs V3 voice | (0.045 + 0.170 + 0.008) x 1,000 + phone $7.65 | $230.65 |
| Retell, ElevenLabs v3 voice, GPT 5.6 Terra | (0.055 + 0.10 + 0.064 + 0.015) x 1,000 | $234.00 |
| Bland, Build plan | $299 + 0.12 x 1,000 | $419.00 |
The spread between the cheapest and dearest standard setups is under $65 a month at this volume. The choice of voice moved the total more than the choice of platform: switching to an ElevenLabs V3 voice roughly tripled the HighLevel figure.
Why do premium voices and models change the price so much?
Because the voice and the language model are charged per minute on top of the platform fee, and premium options cost many times the standard ones. On HighLevel, an ElevenLabs V3 voice is $0.170 a minute against $0.015 for an OpenAI voice. On Retell, model prices range from under $0.01 to $0.32 a minute.
For most business calls, such as booking, answering opening-hours questions or qualifying a lead, a standard voice and a small model do the job. The premium voice is worth testing where the call is long, emotional or high value, such as sales calls or patient intake. Test both on recorded sample calls with real callers' questions before you pay three times as much per minute.
Speech-to-speech models, which hear and answer in audio without converting to text in between, are sold as one bundled rate. HighLevel describes them as lower latency and better at tone. At $0.10 a minute for Gemini Live, they sit between a standard and a premium setup, with no token bill to estimate.
When does a flat monthly plan beat paying per minute?
HighLevel's $97 Unlimited plan beats its own pay-per-use rates once you pass about 1,617 minutes a month with a standard voice, or about 970 minutes against speech-to-speech, before tokens. ElevenLabs' Pro plan beats Creator plus overage at about 1,238 minutes, exactly the minutes Pro includes.
Minutes a month where each flat plan starts to win
The arithmetic, using the published rates:
- HighLevel Unlimited vs pay per use: $97 divided by $0.06 (voice engine plus a $0.015 voice) is 1,617 minutes. Add model tokens to the pay-per-use side and the break-even comes earlier. Unlimited is per location and "subject to fair use", and phone charges apply either way.
- HighLevel Unlimited vs speech-to-speech: $97 divided by $0.10 is 970 minutes.
- ElevenLabs Creator vs Pro: Creator costs $22 plus $0.08 for each minute over 275. That reaches Pro's $99 at 1,237.5 minutes, and Pro includes 1,238. Below that, Creator is cheaper; above it, Pro. Pro also doubles concurrent calls from 10 to 20.
If you already run GoHighLevel, the HighLevel plans have one more advantage that does not show in the price: the agent books into your calendars and writes to contact records without an integration to build. Our GoHighLevel Voice AI page covers what that setup includes.
What costs do AI voice agent price pages leave out?
The per-minute rate leaves out phone numbers and call minutes on some platforms, extra concurrent call lines, HIPAA agreements, transfer minutes, and the work of building, testing and maintaining the agent. For a business, the build is usually the largest single cost.
Another cost these pages leave out is geography. The same HighLevel call costs $68.50 per 1,000 inbound minutes in the US and Canada and $70 in the UK and Australia, but a transfer to a UAE mobile is 21 times the US rate. See AI receptionist running costs by country.
| Cost | What we found on the pricing pages |
|---|---|
| Phone line | Separate on HighLevel ("Phone System charges still apply to all phone calls regardless of plan") and ElevenLabs; listed per minute on Retell and Vapi |
| Concurrent calls | Vapi usage-only includes 4, extra lines $10/month each; Retell includes 20; ElevenLabs from 4 (Free) to 40 (Business) |
| HIPAA | Vapi add-on $2,000/month with a BAA; Retell lists HIPAA/BAA in its compliance table (confirm which plan with Retell); ElevenLabs offers BAAs on Enterprise |
| Transfers to a person | Bland charges transfer minutes ($0.05 or $0.04/min); elsewhere the forwarded leg is billed as telephony |
| Extras | Retell knowledge base +$0.005/min, PII removal +$0.01/min, branded caller ID +$0.10 per outbound call |
| Build and upkeep | Not on any pricing page: call design, knowledge base, calendar and CRM connections, testing, monitoring |
Our own prices for the build, at the time of writing: a small build (2 to 3 automations, such as a voice agent with booking and follow-up) about $1,000, a full setup $2,000 to $5,000, with six months of maintenance included. Our AI receptionist cost guide adds up build and running costs for a receptionist, and the AI receptionist ROI calculator compares them with staff and answering services.
How many calls can an AI voice agent take at the same time?
How many calls run at once depends on the plan's concurrency limit: on the pages we read, Vapi's usage-only plan allows 4, ElevenLabs allows 4 to 40 by plan, and Retell includes 20. Calls beyond the limit wait, fail or cost more, so check it against your busiest hour.
Calls at the same time, by plan
| Platform and plan | Concurrent calls included | Going over |
|---|---|---|
| Retell pay as you go | 20 | "More available on demand" |
| Vapi usage only / Core / Pro | 4 / 10 / 30 | $10 per extra line per month |
| ElevenLabs Free / Creator / Pro / Scale / Business | 4 / 10 / 20 / 30 / 40 | Burst minutes at $0.16, double the standard rate |
A single-location clinic rarely needs more than a few lines at once. A call center or a multi-location business does, and that is where an enterprise plan, or a platform with a high included limit, starts to make sense.
Which AI voice agent platform is cheapest for a small business?
At a few hundred minutes a month, the cheapest usage fees we found were HighLevel Voice AI pay-per-use, Vapi and Retell, all near $0.08 to $0.09 a minute with standard voices. The difference is a few dollars a month, so pick on fit: the tools you already use, call quality and who will maintain it.
Some quick rules from the numbers above:
- Already on GoHighLevel: start with HighLevel Voice AI pay-per-use, and move to Unlimited if minutes pass about 1,600 a month.
- Need full control over the call logic, or several systems: Retell or Vapi, with the agent writing to your CRM through an API.
- Want a predictable bill: an ElevenLabs plan sized to your minutes, or HighLevel Unlimited.
- High-volume outbound with transfers: price Bland's transfer minutes and Build plan against the per-minute platforms.
Our comparison of the best AI voice agent platforms for GoHighLevel covers features and integration as well as price. If you want us to build and run the agent, start with our AI calling agent service.
Questions people ask about AI voice agent pricing
Is GoHighLevel Voice AI cheaper than Retell or Vapi?
On list prices, slightly, for a standard setup: about $0.076 a minute on HighLevel pay-per-use against about $0.09 on Vapi and $0.093 on Retell in our worked example, including the phone line. The gap is small. HighLevel saves more by skipping integration work if your CRM is already GoHighLevel.
Does AI voice agent pricing include the phone number and call minutes?
Not always. Retell and Vapi list telephony per minute, and Vapi includes one free number on its usage plan. HighLevel bills Phone System charges separately on every AI plan. ElevenLabs does not charge for telephony but your provider, such as Twilio, bills you for the line. Always add the phone cost before comparing.
How much does an AI voice agent cost per month for a small business?
For 300 minutes a month, about 75 calls of four minutes, usage fees come to roughly $23 to $28 on the cheapest standard setups we priced, plus the build. A business taking 1,000 minutes pays about $76 to $140. The build and upkeep, from about $1,000 for a small build with us, usually costs more than the first year of minutes.
Do AI voice agents charge for transferring a call to a person?
Bland lists transfer minutes separately: $0.05 a minute on Start and $0.04 on Build. On other platforms the transferred leg is billed as normal phone minutes, so a long transferred call keeps costing per minute until it ends. Check your transfer volume, since an agent that transfers half its calls costs more than its rate suggests.
Are speech-to-speech voice agents more expensive?
Not always. HighLevel's speech-to-speech rate with Gemini Live is $0.10 a minute, bundled with tokens and transcripts, which is more than a standard setup but less than one with a premium voice. OpenAI's realtime model on HighLevel is $0.20 a minute. The advantage is a single predictable rate per minute.
How often do AI voice agent prices change?
Often. HighLevel's voice engine rate took effect May 20, 2026 and its OpenAI and Cartesia voice rates on April 20, 2026, and a voice price change planned for October 13 was cancelled, as our news note reports. Re-check pricing pages before you quote a client or sign a contract.
Related questions and services
- Best AI voice agent platforms for GoHighLevel
- How much does an AI receptionist cost?
- How to build an AI receptionist
- GoHighLevel Voice AI setup
- AI calling agent service
- AI receptionist
Sources and method
Every price was read from the vendor's own page on October 17, 2026, in US dollars, for US calls, without promotional discounts. Where a page shows a range or a calculator, we took the lowest and highest values it displayed. The language model allowance of $0.008 per minute on token-billed platforms is our assumption, explained above. Worked examples are illustrative, not client data.
- HighLevel, AI products pricing (voice engine, TTS, token prices, speech-to-speech, AI Employee plans), read October 17, 2026.
- HighLevel, Phone System pricing and billing guide (US inbound $0.00765/min), read October 17, 2026.
- Retell AI pricing, read October 17, 2026.
- Vapi pricing and Vapi docs, pricing and success packages, read October 17, 2026.
- Bland pricing, read October 17, 2026.
- ElevenLabs Agents pricing, read October 17, 2026.
- Synthflow pricing, read October 17, 2026.
- Twilio US voice pricing (local inbound $0.0085/min), read October 17, 2026.
Vendors change prices without notice. Treat this as a dated snapshot and confirm on the vendor's page before you buy.
