By Service Socket
Retell AI Pricing per Minute (2026): Full Cost Breakdown
Retell AI costs $0.07 to $0.31 per minute. See every component rate, three real per-minute builds, monthly totals, and the costs the calculator leaves out.

Retell AI is a developer platform for building voice agents. It does not sell plans. You pay per minute for each piece of the call, and the total depends on which voice engine, language model, speech voice, and phone carrier you pick.
That makes the headline range, $0.07 to $0.31 per minute, accurate but not very useful. This guide breaks the price into its components, builds three realistic agents, prices them at real call volumes, and lists the costs that never show up in the per-minute calculator.
Every rate below is from Retell's pricing page, checked October 12, 2026. This article is published by Service Socket, which sells a finished AI front office for service businesses. We will say plainly where Retell is the cheaper choice.
Quick answer
- Voice agents: $0.07 to $0.31 per minute on pay-as-you-go, with $10 in free credits and no contract. Enterprise pricing is custom.
- What makes up a minute: Retell voice infrastructure ($0.055), a text-to-speech voice ($0.015 to $0.10), a language model (about $0.002 to $0.32), and telephony ($0.015 through Retell's Twilio numbers in the US, or $0 if you bring your own).
- Typical build: about $0.13 per minute with a mid-tier model, a standard voice, Retell telephony, and a knowledge base.
- Monthly extras: $2 per phone number, $8 per concurrent call beyond the first 20, and $8 per knowledge base beyond the first 10.
- What the calculator leaves out: the time to build, connect, test, and maintain the agent.
Retell AI component pricing (2026)
| Component | Option | Price |
|---|---|---|
| Voice engine | Retell Voice Infra | $0.055/min |
| Text-to-speech | Retell, Cartesia, OpenAI, Minimax, Fish voices | $0.015/min |
| Text-to-speech | ElevenLabs Flash | $0.040/min |
| Text-to-speech | ElevenLabs Multilingual v2 | $0.06/min |
| Language model | GPT 5 mini | $0.008/min |
| Language model | GPT 5 | $0.04/min |
| Language model | Claude 4.5 Sonnet | $0.08/min |
| Language model | Full range on the page | About $0.0016 to $0.32/min; "fast tier" options cost double |
| Telephony | Retell Twilio, United States | $0.015/min |
| Telephony | Your own SIP trunk or carrier | No Retell charge |
Per-minute add-ons: knowledge base +$0.005, advanced denoising +$0.005, safety guardrails +$0.005, PII removal +$0.01, and AI quality assurance $0.10 per minute after the first 100 free minutes.
Monthly and per-event charges: phone numbers $2 per month, Retell SMS $20 per month plus $0.01 per message, concurrency free for the first 20 active calls then $8 per extra concurrent call per month, knowledge bases $8 per month each after 10 free, branded caller ID +$0.10 per outbound call, and batch dialing +$0.005 per dial.
Model names and prices change often. Check the page on the day you build.
Three real per-minute builds
These are example configurations priced from the table above.
| Build | Voice infra | Voice | Model | Telephony | Add-ons | Per minute |
|---|---|---|---|---|---|---|
| Lean | $0.055 | $0.015 (Retell) | $0.008 (GPT 5 mini) | $0.015 | None | $0.093 |
| Typical | $0.055 | $0.015 (Retell) | $0.04 (GPT 5) | $0.015 | $0.005 knowledge base | $0.13 |
| Premium | $0.055 | $0.06 (ElevenLabs Multilingual v2) | $0.08 (Claude 4.5 Sonnet) | $0.015 | $0.005 knowledge base + $0.005 denoising | $0.22 |
The voice engine and telephony are fixed at about $0.07 combined, which is why the range starts there. Your choice of model and voice decides almost everything else. A premium voice and a top-tier model can double or triple the bill for the same call.
What Retell costs per month
Assuming one phone number and fewer than 20 simultaneous calls:
| Monthly minutes | Lean ($0.093) | Typical ($0.13) | Premium ($0.22) |
|---|---|---|---|
| 300 minutes (about 100 three-minute calls) | $29.90 | $41 | $68 |
| 1,000 minutes | $95 | $132 | $222 |
| 2,000 minutes (busy season) | $188 | $262 | $442 |
Each total is minutes x rate + $2 for the phone number. For example, 1,000 x $0.13 + $2 = $132.
On raw usage, that is inexpensive. A 1,000-minute month on the typical build costs less than a single month of most live answering services. If you have someone who can build and maintain the agent, Retell is one of the cheapest ways to put AI on a phone line.
The costs the calculator leaves out
The per-minute rate covers the call. It does not cover turning a call into a booked job.
- Build time. Someone writes the prompt, designs the call flow, connects the calendar or booking system, sets up transfer rules, and tests edge cases: interruptions, wrong addresses, out-of-area callers, emergencies.
- Integrations. Retell offers appointment booking, call transfer, and a knowledge base, plus integrations through tools like Make, n8n, Twilio, and HubSpot. Writing the job into your field-service software usually means webhooks or custom work.
- Maintenance. Prices, service areas, hours, and seasonal offers change. So do the underlying models. Someone has to update the agent and re-test it.
- Monitoring. Retell provides call analytics, transcripts, and optional AI quality assurance at $0.10 per minute. Reviewing them and fixing failures is still a job.
- Everything after the call. Website chat, follow-up texts, quotes, and invoices are separate builds or separate products.
A simple way to price this: estimate your build hours and multiply by what that time costs. If a contractor charges $100 an hour and the build takes 20 hours, that is a hypothetical $2,000 before the first call, plus monthly upkeep. Retell also lists certified solution partners you can hire to build for you; get a quote that separates the build fee from the monthly usage.
Our guide to why customer calls still sound robotic covers the call-quality side of a custom build: latency, talk-over, and corrections.
Retell vs a finished front office
| Retell AI (typical build) | Service Socket Starter | Service Socket Pro | |
|---|---|---|---|
| Price | About $0.13/min + $2/number | $99/month, 500 AI credits | $299/month, 2,500 AI credits |
| Usage beyond the plan | Per minute, every minute | Metered credit overage | Metered credit overage |
| Who builds it | You or a hired partner | Ready to configure; founder-led onboarding | Same |
| Booking | Calendar booking you connect | Availability-aware booking into native leads and jobs | Same, plus schedule board and dispatch suggestions |
| Website chat | Separate chat agent build | Same agent on voice and chat | Same |
| After the call | Webhooks to your systems | Quotes, invoices, online payments | Adds outbound SMS and AI follow-up calls |
| Best for | Teams with a developer, high volume, custom flows | Small shops that want answering and booking working this week | Shops that also want follow-up and dispatch help |
The honest summary: Retell is cheaper per minute. On Service Socket's published credit costs, a three-minute call uses 12 voice credits and a booking adds 5, so Pro's 2,500 credits cover roughly 147 booked three-minute calls before overage. A Retell typical build handles the same calls for about $57 in usage (147 x 3 x $0.13 = $57.33).
What the $299 buys is everything around those minutes: a front office that is already built for HVAC, plumbing, and electrical calls, with booking, chat, emergency warm transfer, quotes and payments, and a team that monitors it. If you would otherwise pay someone to build and maintain those pieces, compare that total, not the per-minute line.
Before deciding, put a number on the calls you are missing today. Our missed-call estimates for plumbing and electrical shops use your own call volume and average ticket.
Who should choose Retell
- You have a developer, or budget for a solution partner, and want full control over prompts, models, and call flows.
- Your volume is high enough that per-minute cost dominates the decision.
- You need custom integrations or outbound campaigns that a packaged product does not offer.
- You are building a voice product of your own, not just answering one shop's phone.
Who should choose a finished front office
- Nobody on your team will own prompts, webhooks, and testing.
- You want booking, website chat, and follow-up working together without separate builds.
- Your volume fits inside a plan, roughly up to a few hundred AI-handled calls a month.
- You want a person to review quality with you instead of reading transcripts alone.
Service Socket's limits, stated plainly: it is not a developer platform, it does not sync jobs into ServiceTitan, Housecall Pro, or Jobber today (those integrations are on the roadmap), and at call-center volume its included credits will not cover everything. For a broader market map, see our guide to AI receptionists for home service. If you are weighing AI against people, our AnswerForce pricing breakdown shows what live answering costs, and our Allo pricing breakdown covers a team phone system with an AI add-on. For four live services priced at the same volume, see answering service cost per minute.
Other options to compare
Our deep dives cover the closest alternatives: Retell AI alternatives for a side-by-side view, Vapi if you want to assemble your own speech and model stack, Bland AI for another developer phone platform, Goodcall for a self-serve AI receptionist with no code, and NextPhone for flat contractor answering. The alternatives hub organizes the full shortlist.
Frequently asked questions
How much does Retell AI cost per minute?
Retell's pay-as-you-go voice agents cost $0.07 to $0.31 per minute. The total is Retell voice infrastructure ($0.055) plus a text-to-speech voice, a language model, and telephony. A typical build with a mid-tier model, a standard voice, Retell telephony, and a knowledge base comes to about $0.13 per minute.
Does Retell AI have a monthly fee?
There is no platform subscription on pay-as-you-go. Monthly charges apply to phone numbers ($2 each), concurrency beyond 20 active calls ($8 each), knowledge bases beyond 10 ($8 each), and Retell SMS ($20). Enterprise pricing is custom.
Is there a free trial?
Retell gives new accounts $10 in free credits with full platform access and no commitment.
Why does Retell show a price range?
The minute price depends on the components you choose. The voice engine and telephony are about $0.07 together, and the language model and voice you add decide the rest. Premium models and voices can push a minute past $0.30.
Is Retell AI good for small service businesses?
It is inexpensive per minute, but it is a building block, not a finished receptionist. Small shops without a developer often spend more on setup and upkeep than on minutes. A finished AI front office trades a higher monthly price for booking, chat, and monitoring that already work.
What is cheaper: Retell or Service Socket?
Retell is cheaper per minute. Service Socket costs more per call but includes a built front office: booking into native jobs, website chat, emergency warm transfer, quotes and payments, and founder-led onboarding. Compare Retell's usage plus your build and maintenance costs against Service Socket's plan price.
Price the whole build, not just the minute
Retell's pricing is transparent and genuinely low. The decision is whether you want to buy minutes and build the receptionist yourself, or buy a receptionist that is already built.
If you want the second, compare the answering-service alternatives, review Service Socket pricing, or get started without writing a prompt.