Vapi vs Retell AI 2026: Real Cost per Minute, Compared

Short answer: Retell AI is the faster, simpler buy for a team shipping its first phone agent: $0.07 to $0.31 a minute with voice infrastructure bundled and 20 concurrent calls free. Vapi wins on control and can win on price: $0.05 a minute for hosting, with speech, model and voice passed through at cost, about $82 to $129 per 1,000 minutes on its own calculator. Verified on both pricing pages, October 2026.

Vapi vs Retell AI at a glance

Sources: Vapi pricing and Retell AI pricing, both read 28 October 2026. Prices in US dollars before tax.

Picture a support team pricing its first AI phone agent: an inbound line that answers order questions, about 1,000 minutes a month at launch and 10,000 if it works. The sticker price is not the bill. What you pay depends on the model, the voice, the phone carrier and how many calls ring at once, and the two platforms bill those parts differently.

How much does Vapi cost per minute?

Vapi charges $0.05 a minute for hosting, prepaid, and passes everything else through. Its pricing FAQ lists four per-minute components on top of hosting: speech-to-text, the language model, text-to-speech and transport, "each billed at the provider's listed price with no markup."

Vapi's calculator at 1,000 minutes shows the spread:

Source: Vapi pricing calculator, read 28 October 2026.

Transport is free on Vapi's own telephony, SIP, WebSockets and Daily WebRTC. If you bring a carrier, Vapi lists Twilio at $0.008 a minute inbound and $0.014 outbound, Vonage at $0.00814 and Telnyx at $0.0055, charged by that provider rather than by Vapi. The model choice moves the bill most: the gap between Vapi's cheapest and priciest OpenAI option is $37 per 1,000 minutes, more than the transcriber and the voice combined.

Vapi's packages sit on top of usage. Without one you get 4 concurrent calls, 1 phone number and 14 days of call data retention. Core at $29 a month raises that to 10 concurrent calls, 5 numbers and 30 days, and adds zero data retention. Pro charges 10% of your hosting fees with a $999 monthly minimum for 30 concurrent calls, a Slack channel and a 99% uptime SLA. Extra call lines cost $10 each a month on any package.

How much does Retell AI cost per minute?

Retell prices voice agents at $0.07 to $0.31 a minute on pay-as-you-go. Every minute carries Retell's voice infrastructure fee of $0.055, then you add a model, a voice and telephony from its price list.

Source: Retell AI pricing, read 28 October 2026.

Retell's own calculator opens at $0.11 a minute: $0.04 for the model, $0.055 voice infra, $0.015 voice and no telephony fee (read 30 October 2026). Billing rounds to the nearest second with no per-call rounding, failed calls are free, and once a call transfers to a human the agent fee stops and only telephony continues. That last point matters for a support line that hands off a share of calls.

Retell's monthly items are small but real: $2 per Retell phone number, $8 per concurrent call beyond the 20 included, and $8 per knowledge base after 10 free ones. Retell lists a verified phone number at $10 in one table and as a one-time $10 fee in another, so check that line at checkout.

Which is cheaper at 1,000 and 10,000 minutes a month?

Here is the same inbound support line priced both ways. The Vapi rows use its calculator's component ranges; the Retell rows use two builds from its price list, one budget model and one mid-tier model, each with a $0.015 platform voice and Retell's US telephony. All totals are our arithmetic on prices read 28 October 2026.

Dupple data: at 10,000 minutes a month, Vapi on its own telephony costs $818 to $1,289 against $1,092 for Retell's budget build and $1,492 for its Claude 5.5 Sonnet build, so Vapi's pass-through beats Retell's bundle whenever you pair it with a cheap model (our arithmetic, October 2026).

Two fixed costs can flip the result. Concurrency is the first: a busy support line rarely stays at 4 simultaneous calls. Retell includes 20 at no charge; on Vapi, 20 lines means Core plus 10 extra lines, $129 a month by our arithmetic, which erases most of Vapi's edge at 1,000 minutes. Compliance is the second: Vapi's HIPAA add-on costs $2,000 a month per organization, more than the whole usage bill in every row above.

The model is the lever on both. Moving from a budget model to a mid-tier one adds $0.04 a minute on Retell and up to $0.0375 on Vapi's OpenAI range, so pick the cheapest model that handles your call script before comparing platforms.

How far do the free credits go?

Retell's $10 covers about 92 minutes on the budget build above, and Vapi's $5 covers roughly 39 to 61 minutes at its calculator's range (our arithmetic). Either is enough for a few dozen test calls, not for a pilot with real customers. Retell's extra 20 concurrent calls also mean a load test costs nothing beyond minutes.

Do I need a developer for Vapi?

For anything past a demo, realistically yes. Vapi sells itself on control: you own and operate your agents, models and data, choose each provider, and work through its API and separate organizations for dev, staging and production. That flexibility is the reason its per-minute price can drop, and it assumes someone on the team can wire providers, carriers and tools together.

Retell is closer to ready-made. Its pay-as-you-go plan includes pre-built agent templates, simulation testing, call analytics, webhooks and API access, and a dashboard calculator that prices your build as you click. A support or ops lead can stand up a Retell agent without an engineer; a Vapi build usually needs one for the integrations that make the agent useful.

Which handles HIPAA and compliance?

Both can, at very different price points. Vapi treats HIPAA as an add-on: $2,000 a month per organization, with a standard data processing agreement and business associate agreement. Zero data retention comes with any Success Package, starting with Core at $29.

Retell's comparison table ticks HIPAA/BAA, PII redaction, opt-out recording and custom data retention on pay-as-you-go, and its page describes the platform as SOC 2 certified and HIPAA-ready. Custom BAA, MSA and DPA terms, SSO and role-based access control are Enterprise items with custom pricing. For a clinic or health startup, Retell's entry point for HIPAA is far cheaper on the published pages; confirm the BAA terms with Retell before routing patient calls.

Strengths and weaknesses

Vapi

• Strength: $0.05 a minute hosting with every provider passed through at cost, so cheap models make cheap minutes.
• Strength: free transport on Vapi telephony, SIP, WebSockets and WebRTC.
• Strength: zero data retention from the $29 Core package.
• Weakness: only 4 concurrent calls without a package, and every extra line adds a monthly fee.
• Weakness: HIPAA costs $2,000 a month per organization.
• Weakness: the Pro package carries a $999 monthly minimum.

Retell AI

• Strength: 20 concurrent calls and 10 knowledge bases included at no monthly fee.
• Strength: per-second billing, no charge for failed calls, and the agent fee stops after a transfer.
• Strength: HIPAA/BAA listed on pay-as-you-go.
• Weakness: the $0.055 voice infra fee applies to every minute, whatever model you choose.
• Weakness: ElevenLabs voices cost $0.040 to $0.10 a minute on Retell's list.
• Weakness: SSO, custom contract terms and 24/7 support need Enterprise.

Who should pick which

• Support or ops team, no engineers, launching in weeks: Retell AI on pay-as-you-go with a budget model.
• Engineering team at 10,000+ minutes that tunes every component: Vapi on Core, paired with the cheapest model that passes your call tests.
• Healthcare team with patient calls: Retell, after confirming its BAA; Vapi's HIPAA add-on only makes sense at large volume.
• Spiky call volume: Retell, because 20 concurrent lines cost nothing extra.

Alternatives to Vapi and Retell

If you want a phone agent without building one, our ranking of the best AI receptionists covers done-for-you services, and how much an AI receptionist costs prices them per call. For the full field of platforms, see the best AI voice agents; for the voice layer alone, the best AI voice generators. Teams replacing a human phone queue rather than adding to it can compare call center software, and if a text bot would do the job, how much an AI chatbot costs is the cheaper path. Our guide to building an AI agent covers the groundwork either platform assumes, and the best AI customer support tools cover chat and email for the same queue. Engineers testing agents before launch can also compare AI agent observability tools.

Toolradar data: Toolradar, our software directory, tracks 132 AI voice tools as of 27 October 2026, and 70% of them are free, freemium or open source (Toolradar AI voice ranking). Both platforms have profiles there: Vapi on Toolradar and Retell AI on Toolradar.

How we compared

Both pricing pages and their calculators were read on 28 October 2026, with the same inbound scenario on each: US calls, a $0.015 platform voice on Retell, and Vapi's default Deepgram, OpenAI and ElevenLabs ranges. Vapi's calculator shows model and voice prices as ranges rather than per model, so its rows are ranges. Totals are our arithmetic, before tax. We did not build or test agents on either platform; features are described from the vendors' own pages.

FAQ

Bottom line

Retell AI is the better default for most teams that use AI at work: it prices a working phone agent in one number, includes the concurrency a support line needs, and lists HIPAA without a four-figure add-on. Vapi is the better tool for engineers who will tune the model, voice and carrier, and at scale with a cheap model it delivers the lower per-minute bill. Price your own script on both calculators before you commit, because the model you pick moves the total more than the platform does.

Cite this: Dupple, "Vapi vs Retell AI 2026: Real Cost per Minute, Compared", November 2026.

Building AI into your team's work? Start the Dupple X trial for the workflows and prompts that turn AI tools into results.