Retell AI vs Vapi (2026): Per-Minute Pricing, Concurrency and Which to Build On

Retell AI vs Vapi, in short: both are developer platforms for AI phone agents billed by the minute. Vapi charges a $0.05 platform fee and passes speech, model, voice and phone costs through at cost; Retell charges $0.055 for voice infrastructure and publishes a rate card for every other part, landing at $0.07 to $0.31 a minute. A typical build costs about $0.11 to $0.13 a minute on either. Retell includes 20 concurrent calls; Vapi's base tier includes 4.

More on this topic: Calls, SMS and AI voice agents, every guide in one place.

At the per-minute level the two are close enough that your choice of model and voice matters more than the platform. The differences that decide it are concurrency, compliance pricing, and how much of the stack you want to choose yourself.

Key takeaways

Is Retell AI or Vapi cheaper?

Neither platform is meaningfully cheaper per minute once you add the model, voice and phone line. The fixed costs differ more: concurrency, subscriptions and compliance.

A mid-range build — a $0.04 model, a standard voice and an outbound phone line — is about $0.125 a minute on Retell and about $0.11 to $0.14 on Vapi. At 3,000 minutes a month that is roughly $330 to $420 on either. The per-component breakdown, and monthly totals at other volumes, are in AI voice agent cost per minute; for a bundled-rate alternative, see Bland AI vs Retell AI.

Pricing for both products was checked against the public Scalry and Retell AI and Vapi pricing pages in October 2026. SaaS pricing changes often — confirm the current numbers with each vendor before you commit to a plan.

Retell AI vs Vapi: how do the features compare?

Retell bundles more on one rate card and includes more concurrency; Vapi gives you each provider at cost and finer control of the stack.

The costs that don't appear on either pricing page

• Silence and hold time are billed on Retell, and on any per-minute platform the agent's time on voicemail counts.
• Concurrency is the fixed fee that surprises outbound teams: a 30-line campaign is $80 a month extra on Retell and needs Pro or extra lines on Vapi.
• Compliance: Vapi's HIPAA add-on is $2,000 a month; get Retell's terms in writing.
• Premium voices and top models can more than double the minute rate on either platform.
• Engineering time to build, test and maintain the agent is the largest cost for most small teams.

Where Retell AI wins

More concurrency without a plan

Twenty simultaneous calls are included on pay-as-you-go. On Vapi the free tier includes four, and getting to thirty means the Pro plan or extra lines at $10 each.

A rate card you can budget from

Every component has a published per-minute price, including dozens of models and voices, so a finance team can forecast the bill before the first call.

Built-in extras

Knowledge bases, batch calling, branded caller ID, SMS and speech-to-speech models are priced on the same page and work without wiring in more providers.

Where Vapi wins

A slightly lower platform fee

$0.05 a minute against Retell's $0.055, with providers at cost. At high volume and with cheap providers, the half-cent adds up.

Full control of the stack

Choose your own transcriber, model, voice and transport, including free Vapi telephony and WebRTC. Teams that already have provider contracts can use them.

Clear data-retention tiers

Retention of raw data is stated per plan — 14, 30 or 180 days, or custom — with zero data retention on any paid package.

Retell AI vs Vapi: which should you choose?

Choose Retell for concurrency and a predictable rate card; choose Vapi for provider control at low concurrency.

Run the same agent on both for a week with your real model and voice. The per-minute numbers will be close; the deciding factors will be concurrency, how quickly your team builds on each, and the compliance terms you need. If neither fits, see Vapi alternatives for bundled and managed options.

How to test Retell AI and Vapi without wasting a fortnight

Feature matrices settle less than two days of real use. Both products will demo well; what you want to know is which one is annoying on a Tuesday.

• Rebuild one real campaign in each, end to end — the page, the form, the follow-up, the handoff. Not a test campaign; the one that matters.
• Import a real segment, not ten sample contacts. Field mapping problems only appear with real data.
• Time the boring tasks. Logging an activity, editing a sequence, finding last month's numbers. You will do these hundreds of times.
• Give it to the person who will actually use it, not the person evaluating it. Adoption is decided by them.
• Price both at next year's size before deciding — Retell AI and Vapi scale on different axes, so today's comparison will not hold.
• Test the export. Confirm you can get your data out on day one, not on the day you want to leave.

Questions to ask before you commit

Whichever way you are leaning, these are the questions that change the answer. Work through them against your own numbers rather than anyone's feature matrix — including this one.

• How many calls at once at peak? That sets the concurrency bill.
• Which model and voice? They move the minute rate more than the platform.
• Do you need HIPAA? Get pricing in writing.
• Who will build and maintain the agent? Both are developer platforms.
• Where do call outcomes go? Plan the CRM side before launch.

If you cannot answer the first three, you are not ready to choose between Scalry and Retell AI or Vapi yet — you are still deciding what you need, which is a cheaper problem to solve first.

A third option, briefly

If you would rather not build a voice agent at all, business platforms bundle one with the CRM. Scalry's voice agents are set up from templates with calling windows, a calls-per-minute rate and do-not-call suppression, and outcomes land on the contact record; call time draws on the plan's credits. You give up the model-level control that Retell and Vapi offer. For what a voice agent should and should not do on a sales call, see AI voice agents for lead qualification.

That is the whole pitch, and it is last for a reason — if one of the two tools above fits, use it. Our pricing study compares thirteen platforms including both, and the cost calculator prices them against your own team size.

Frequently asked questions

Is Retell AI cheaper than Vapi?

Not meaningfully per minute. Vapi's platform fee is $0.05 against Retell's $0.055, but a typical build costs about $0.11 to $0.13 a minute on either once model, voice and phone are added. Retell includes more concurrency for free.

How much does Vapi cost per minute?

Vapi charges $0.05 a minute for hosting, plus transcription, model, voice and telephony at cost. Its own estimate for 1,000 minutes a month is $82 to $129.

How much does Retell AI cost per minute?

Retell lists $0.07 to $0.31 a minute: $0.055 for voice infrastructure, $0.015 for telephony, and the chosen voice and model on top.

How many concurrent calls do Retell and Vapi include?

Retell includes 20, then $8 per extra line a month. Vapi includes 4 on its free tier, 10 on Core and 30 on Pro, with extra lines at $10 a month.

Do Retell AI and Vapi support HIPAA?

Vapi offers a HIPAA add-on at $2,000 a month with a BAA. Retell lists HIPAA and BAA among its compliance features but does not publish a price; ask for terms in writing.

Which is easier to start with, Retell or Vapi?

Both are developer platforms. Retell bundles more features on one rate card, which usually makes the first working agent quicker; Vapi gives more control over each provider.