A missed call at a clinic, salon or service business is often a missed booking. That is why owners now ask what an AI phone agent costs. The honest answer: AI voice agent cost is a per-minute bill made of five parts, and the published rates on Retell, Vapi and Bland only make sense once you see what each rate includes.
Below: prices checked on each vendor’s pricing page on 22 October 2026, a monthly AI receptionist cost example, and when a custom build pays off.
Key takeaways
- As of 22 Oct 2026, Retell lists $0.07–$0.31 per minute, Vapi lists $0.05 per minute for hosting plus pass-through model costs, and Bland lists $0.14 per minute on its no-fee Start plan.
- The headline rate is rarely the full bill. Check whether telephony, the LLM, speech-to-text and the voice are included or charged on top.
- At 1,000 minutes a month, our worked example lands at roughly $90–$150 in usage fees on any of the three platforms, before setup and integration work.
- A custom build makes sense when the agent must book into your own systems, speak Hindi or Arabic well, keep data in a set region, or run at volumes where per-minute margins add up.
In this guide
- What drives AI voice agent cost per minute
- Retell vs Vapi vs Bland: published prices
- AI receptionist cost for a small business: worked example
- When a custom AI voice agent build pays off
- Hidden costs and risks
- FAQ
What drives AI voice agent cost per minute
Every minute an AI voice agent spends on a call pays for five layers: the phone line, speech-to-text, a language model (LLM) that decides what to say, text-to-speech for the voice, and the platform or servers that tie them together in real time.
Bland’s per-minute rate bundles the LLM, speech-to-text and text-to-speech; telephony is separate. Retell and Vapi charge a platform fee and add each component, so your model and voice choice moves the price. Telephony itself is cheap: Twilio lists $0.0085 per minute to receive calls on a US local number, plus $1.15 a month for the number.
Retell vs Vapi vs Bland: AI voice agent cost per minute
Here is what each platform publishes. All three are pay-as-you-go to start, with enterprise tiers priced on request.
| Platform | Published rate | What it includes | Notable limits and fees |
|---|---|---|---|
| Retell AI | $0.07–$0.31/min pay-as-you-go | $0.055/min voice infrastructure, plus TTS (from $0.015/min), LLM (e.g. $0.024/min for GPT 5.4 mini) and telephony ($0.015/min via Retell’s Twilio) | 20 concurrent calls included, then $8 per call/month; numbers $2/month; billed per second |
| Vapi | $0.05/min hosting, providers at cost | Transcriber (Deepgram $0.0095–$0.0099/min), model (OpenAI $0.0077–$0.0452/min), voice (ElevenLabs $0.0146–$0.0238/min) | Telephony charged by the provider; 4 concurrent calls without a package; Pro package $999/month minimum |
| Bland AI Start | $0.14/min, $0 platform fee | LLM, speech-to-text and text-to-speech; no token charges | Telephony separate; $0.05/min transfer time; 10 concurrent calls, 100 calls/day |
| Bland AI Build | $0.12/min + $299/month | Same as Start | 50 concurrent calls, 2,000 calls/day |
On Retell vs Vapi, the difference is less about the base fee than about defaults. Retell includes 20 concurrent calls and one clear component price list. Vapi has the lower platform fee and passes model costs through at cost, but you arrange telephony yourself.
AI receptionist cost for a small business: a worked example
Say a clinic or home-services firm routes 1,000 minutes of inbound calls a month to an AI receptionist on a US local number. Using only the published rates above, the monthly usage bill looks like this:
| Option | How we priced one minute | Approx. per minute | Approx. per month (1,000 min) |
|---|---|---|---|
| Retell | $0.055 infra + $0.015 voice + $0.024–$0.064 LLM (GPT 5.4 mini to Claude 5 Sonnet) + $0.015 telephony | $0.11–$0.15 | $111–$151 incl. $2 number |
| Vapi | $0.05 hosting + Deepgram + OpenAI + ElevenLabs ranges + $0.0085 Twilio inbound | $0.09–$0.14 | $91–$139 incl. $1.15 Twilio number |
| Bland Start | $0.14 bundled + $0.0085 Twilio inbound | $0.15 | About $150 incl. $1.15 Twilio number |
| Bland Build | $0.12 bundled + $0.0085 Twilio inbound + $299 fee | $0.13 + fee | About $429 incl. $1.15 Twilio number |
At small volumes the three cost about the same, so pick on features, voice quality and limits. Bland’s Build plan only beats its Start plan once its $0.02 per-minute saving covers the $299 fee, at roughly 15,000 minutes a month.
The table leaves out what is often bigger than the usage bill: call flows, calendar or CRM integration, testing and hand-off to staff. Our guide on the cost to build an AI agent explains how that build work is priced.
Planning an AI receptionist or phone agent? Send us your call volume, the systems it must book into and the languages your callers speak. Eoxys will tell you whether a platform setup is enough or a custom build pays off. Get a real project estimate within 24 hours. NDA on request.
When a custom AI voice agent build pays off
A platform is enough when the agent answers FAQs, takes messages, books into a supported calendar and transfers calls. A custom build, usually on top of the same speech and model APIs, starts to pay off when:
- The agent must act inside your systems. Real-time slot checks in a clinic or salon booking system, order lookups or CRM updates need dependable integrations, error handling and audit logs.
- Language quality matters. Hindi, Arabic or mixed-language callers need careful voice, transcription and prompt testing. ElevenLabs lists 90+ languages for its v4 voices, but support on a list is not the same as sounding natural to your callers.
- Data has to stay in a region or on your servers. Bland lists data residency and on-prem or VPC deployment on its Enterprise tier; with your own stack you control this from day one. Indian businesses should also plan for the DPDP Act, covered in our DPDP compliance guide for AI apps.
- Volume is high. Your own components can cost less per minute, but you own hosting, monitoring and upkeep.
For a sense of the floor, a self-built US inbound stack pays Twilio $0.0085/min to receive the call and $0.0044/min for Media Streams to stream audio to your server, ElevenLabs $0.39 per hour (about $0.0065/min) for realtime speech-to-text and $0.04 per 1,000 characters for Flash/Turbo speech (as of 22 Oct 2026). Add LLM tokens and servers on top.
| Build path | Typical scope | Effort (estimate) |
|---|---|---|
| Platform setup | Prompt and call flow, knowledge base, call transfer, calendar link, testing | Roughly 20–60 hours |
| Platform plus custom integrations | Booking or CRM APIs via webhooks, post-call summaries, dashboards, multi-location routing | Roughly 80–200 hours |
| Fully custom voice agent | Own orchestration, telephony streaming, model routing, language tuning, regional hosting, monitoring | Roughly 300–700+ hours |
If you are weighing the two routes more broadly, our custom AI agent vs off-the-shelf tools guide walks through the decision. Eoxys builds both kinds of project through its chatbot and AI agent development and AI and machine learning teams.
Hidden costs and risks
- Silence is billed. Retell’s pricing FAQ says the whole call is billed, including silence and hold time, because transcription stays active. Long holds cost money on any per-minute plan.
- Transfers and concurrency. Bland charges transfer minutes; Retell stops its AI fee after a transfer but telephony continues. Concurrency caps (4 on Vapi without a package, 10 on Bland Start) matter at peak hours.
- Model choice swings the bill. On Retell’s list, LLM fees run from $0.0016 to $0.64 per minute (fast tier). Use the smallest model that passes your tests.
- Compliance tiers. HIPAA business associate agreements sit on enterprise tiers or add-ons: Bland lists a BAA on Enterprise and Vapi sells HIPAA as an add-on. Healthcare buyers should budget for this.
- Accuracy. Voice agents can mishear names, numbers and accents. Test real call scenarios and keep a clear hand-off to a person.
Voice is one channel. If you also run chat support, see how AI chatbots are changing customer support for the same patterns in text.
FAQ
How much does an AI receptionist cost per month?
Usage fees depend on call minutes. Using published prices as of 22 October 2026, about 1,000 inbound minutes a month works out to roughly $90–$150 on Retell, Vapi or Bland’s Start plan, including a phone number. Setup, integrations with your booking system and ongoing tuning are extra and often cost more than the minutes.
What is the AI voice agent cost per minute?
As of 22 October 2026, Retell lists $0.07–$0.31 per minute, Vapi charges $0.05 per minute for hosting plus speech, model and voice costs at provider rates, and Bland charges $0.14 per minute on Start or $0.12 on Build with a $299 monthly fee. Telephony is usually added on top.
Retell vs Vapi: which is cheaper?
At small volumes they are close. In our 1,000-minute example, Retell came to about $0.11–$0.15 per minute and Vapi to about $0.09–$0.14, depending on the model and voice. Retell includes 20 concurrent calls; Vapi includes four without a package. Choose on features, voice quality and support, then compare real invoices.
How do I get an estimate from an AI voice agent development company?
Send Eoxys your monthly call volume, the systems the agent must connect to, the languages your callers speak and any data-location rules. We will recommend a platform setup or a custom build and send a real project estimate within 24 hours, with fixed-cost, dedicated-team or hourly options. NDA on request.
Get your AI voice agent built by Eoxys
Eoxys IT Solution has built software from Jaipur since 2009, with 700+ projects delivered across web and mobile apps, e-commerce, fintech, health and EdTech. See the work on our portfolio, or hire dedicated AI developers for a longer build.
Whether you need an AI receptionist on a platform or a custom voice agent wired into your booking system and CRM, get a real project estimate within 24 hours. NDA on request.
Related reading:
- How Much Does It Cost to Build an AI Agent in 2026?
- Custom AI Agent vs Off-the-Shelf Tools: How to Decide
- How AI Chatbots are Transforming Customer Support for Businesses
Sources
Checked on 22 Oct 2026 against the official pricing pages of Retell AI, Vapi, Bland AI, Twilio Programmable Voice (US) and the ElevenLabs API. Vapi’s Core package price is shown only through an interactive calculator, so it is not quoted. Monthly figures are our arithmetic from these published rates; build hours are estimates, not quotes.
- Retell AI: Pricing
- Vapi: Pricing
- Bland AI: Pricing
- Twilio: Programmable Voice pricing, United States
- ElevenLabs: API pricing
Shafeeq Khan is a web developer who creates modern, responsive, and user-friendly websites with clean code and creative design
