Retell AI Pricing vs Vapi Pricing on the All-In Cost Per Minute

Retell AI pricing starts at $0.055 a minute for infrastructure and lands at $0.133 to $0.245 a minute all in, depending on the model. Vapi pricing starts at $0.05 a minute for hosting and lands at about $0.090 to $0.143 all in once transcription, model, voice and telephony are added. Minute for minute, Vapi is about 4 cents cheaper on comparable settings. Retell wins below roughly 3,000 minutes a month for any team that needs 20 simultaneous calls, because Retell includes 20 concurrent lines and Vapi charges for every line past its plan. Both numbers below come straight from the two pricing pages, added up line by line.

How much does Retell AI cost per minute?

Retell bills four lines on every call minute: $0.055 for its voice infrastructure, $0.015 for a platform voice ($0.040 for an ElevenLabs voice), the model you pick, and $0.015 for US telephony. The model is the line that moves. The cheapest model on the page today is Gemini 3.5 Flash at $0.048 a minute. GPT 4.1, which many older comparisons still quote at $0.045, is no longer listed. Retell states its own range as $0.07 to $0.31 a minute, with no minimum and $10 in starting credit.

The extras matter more than people expect. Branded caller ID adds $0.10 per outbound call, which on a three minute call is another $0.033 a minute. Batch dialing adds $0.005 a dial. Verified phone numbers are $10 a month each, knowledge bases $8 a month after the first ten, and SMS $0.01 a message. None of that shows up in the headline number.

How much does Vapi cost per minute?

Vapi charges $0.05 a minute for hosting and passes every provider through at cost. On its own pricing page that is $0.0095 to $0.0099 for Deepgram transcription, $0.0077 to $0.0452 for an OpenAI model and $0.0146 to $0.0238 for an ElevenLabs voice. Add US telephony at $0.008 to $0.014 a minute and the all-in minute runs from about $0.090 with a small model to $0.143 with the top of each range. Vapi gives $5 in starting credit.

The fixed side is where Vapi charges. Pay as you go includes 4 concurrent calls and 1 phone number. The Core package is $29 a month for 10 concurrent calls and 5 numbers. Pro is a $999 a month minimum, billed as 10 percent of hosting, for 30 concurrent calls and a 99 percent uptime SLA. Extra lines are $10 each a month, and HIPAA compliance is a $2,000 a month add-on.

Is Vapi cheaper than Retell?

Per minute, yes, by about 4 cents on comparable settings. On the cheapest sensible build the gap is $0.090 on Vapi against $0.133 on Retell, $0.043 a minute. On a premium build (the top OpenAI tier and an ElevenLabs voice on Vapi, GPT 5.4 and an ElevenLabs voice on Retell) it is $0.143 against $0.190, $0.047. Per month, it depends on how many calls ring at once, because that is where Vapi collects its fees.

The $129 is Core at $29 for 10 lines plus 10 more at $10. Divide it by the $0.043 gap and you get the break-even: 3,000 minutes a month. Below that, a team that needs 20 lines pays less on Retell, which includes 20 concurrent calls and charges $8 a line only past that. Above it, Vapi. A team that needs just 10 lines crosses much earlier, at about 674 minutes, and a single-line agent on Vapi pay as you go is cheaper from the first minute. Vapi Pro at $999 needs about 23,200 minutes a month of the same gap to pay for itself against Retell.

What the Retell vs Vapi price per minute leaves out

Engineering time. Vapi's low minute comes from letting you pick and key every provider yourself, which is the right design for a team with a developer and the wrong one for a team without. By our estimate, wiring and tuning the providers on Vapi is about 10 to 20 more hours than the same agent on Retell. At the $60 to $150 an hour marketplace developers charge, that is $600 to $3,000 up front, and at a $0.043 gap it takes 14,000 to 70,000 minutes to earn back. A small business taking 1,500 minutes of calls a month would need somewhere between nine months and four years.

Compliance. If the agent handles patient calls, Vapi publishes a HIPAA add-on at $2,000 a month. Retell offers a BAA and custom MSA and DPA terms on its Enterprise plan, which has no published price. Neither is a small-clinic purchase, and anyone quoting "Retell is HIPAA ready at $0.07" is skipping the Enterprise step.

Uptime of the thing you own. Both platforms call your server mid-conversation to look up a customer, check a calendar or write to a CRM. If that endpoint is slow or down, the agent stalls on a live caller, and neither vendor's SLA covers your code. Whoever builds the agent should put simple uptime monitoring on every webhook URL before the first real call, because the first person to notice an outage should not be a customer on hold.

Which is better, Vapi or Retell?

Retell is better for a business that wants a managed builder, predictable itemized billing and up to 20 simultaneous calls without a platform fee: an answering line, an appointment setter, a lead-qualifying agent for a single office. Vapi is better for a team with an engineer on hand, high and steady volume past about 3,000 minutes, or a need to bring its own models and carriers. The shortlist rarely stops at these two; our Vapi alternatives page prices Telnyx, LiveKit, ElevenLabs and Bland AI the same way, and the Synthflow alternatives post covers what changed when Synthflow moved to a $30,000 a year minimum.

Retell AI pricing for a real use case

Take a real estate team answering 200 new leads a month with a three minute qualifying call each: 600 minutes. On Retell with Gemini 3.5 Flash that is $79.80 a month, and 12 follow-up texts per lead add $24. On Vapi pay as you go at $0.090 it is $54, but the 4-line limit will bite on a Monday morning after a weekend of portal leads. The platform bill is small either way. The real cost is the build that connects the agent to the CRM and calendar, covered in our AI agent for real estate pricing, and for general call answering the AI receptionist comparison.

Does Retell AI charge for concurrency?

Not for the first 20 concurrent calls. Retell includes 20 simultaneous calls on pay as you go and charges $8 a month for each extra line. Vapi includes 4 on pay as you go, 10 on the $29 Core package and 30 on the $999 Pro package, with extra lines at $10 each. For bursty inbound traffic, Retell's included 20 lines are worth up to $129 a month against Vapi's Core package.

How to get a Retell or Vapi agent built without overpaying

Price the build in hours, not in platform. A single-purpose phone agent that answers, qualifies and books into one calendar is typically 40 to 100 hours; one that writes to a CRM and hands off to people is 80 to 140. Our AI voice agent development page lays out what each part costs, and the AI agent development cost audit shows how far published quotes drift from the hours underneath them.

Then describe the agent once in the hire brief at the top of this page: the calls it takes, the systems it touches, how many can ring at once. The brief comes back with an effort band and matched developers who have shipped on Retell or Vapi, with their rates visible. Buyers on the $99 a month Growth plan pay no hire fee on top.

The short version

Retell AI pricing is $0.133 to $0.245 a minute all in, with 20 concurrent calls included. Vapi pricing is $0.090 to $0.143 a minute all in, with 4 to 30 lines depending on the package and $10 for each one after. Vapi is cheaper per minute, Retell is cheaper for a busy line below about 3,000 minutes a month, and on both the developer's hours cost more than the first year of minutes.