A8gent
A8gent
·
Comparison

Vapi vs Retell AI: Real Per-Minute Cost

Vapi vs Retell AI on August 2026 published pricing: unbundled $0.05/min orchestration against an itemised $0.07-$0.31 all-in rate, and who each suits.

Vapi vs Retell AI: Real Per-Minute Cost visual
AI Agent Starter Kit cover

Free download

AI Agent Starter Kit

Workflow picker, ROI worksheet, 40-prompt pack, and the first-agent rollout playbook. The 1-hour version of the entire A8gent system.

Workflow picker
ROI worksheet
Prompt checklist
Team rollout plan
Open it now

No spam. Worksheet + prompt checklist + rollout plan delivered instantly.

Contents
  1. Fast answer
  2. Quick facts
  3. What this covers
  4. Why it matters
  5. Where to go next
  6. Outcomes
  7. Step by step
  8. Pick Vapi when, pick Retell AI when
  9. The model choice matters more than the platform choice
  10. Where both of these are the wrong answer
  11. At a glance
  12. Mistakes to avoid
  13. FAQ
  14. Related guides
TL;DRAnswer-first

Fast answer

Choose Vapi when you have someone who writes code and wants to own the voice stack, and Retell when you want an accurate forecast before you build. As of August 2026 Vapi charges $0.05 a minute for its own orchestration and passes speech-to-text, model and speech synthesis through at provider cost - or at nothing to Vapi if you bring your own keys - so 1,500 minutes a month is $75 to Vapi plus your providers, typically landing between $120 and $200 all-in. Retell publishes the most itemised pricing in the category: $0.055 a minute for voice infrastructure, $0.015 for text-to-speech, $0.003 to $0.16 for the language model and $0.015 for US telephony, giving an all-in band of $0.07 to $0.31 and roughly $105 to $465 for the same 1,500 minutes. Neither is cheap or expensive as a rule - the spread inside Retell's own range is wider than the gap between the two products, and it is driven almost entirely by which language model you pick. The honest split is not price, it is assembly: Vapi hands you components, Retell hands you a configured pipeline plus a visual builder.

Facts last verified 2026-08

Quick facts

Vapi platform fee
$0.05/min plus separately billed providers (as of Aug 2026)
Retell all-in range
$0.07-$0.31/min, itemised by component (as of Aug 2026)
Reference workload
500 calls a month at 3 minutes each = 1,500 billable minutes
Vapi at 1,500 min
$75 to Vapi, typically $120-$200 all-in with providers
Retell at 1,500 min
$105-$465, driven almost entirely by model choice
Concurrency included
Vapi 10 lines, Retell 20 calls
Compliance
Vapi HIPAA $2,000/mo and ZDR $1,000/mo; Retell PII removal $0.01/min
Best for
Vapi for developer control, Retell for a forecastable bill

On this page

What this page covers

A comparison visitor should understand the tradeoff, the best-fit scenario, and the next diagnostic tool to confirm the choice.

  1. 01Quick verdict
  2. 02Pricing
  3. 03Who maintains it
  4. 04Concurrency
  5. 05Decision table
  6. 06Where both are wrong

Why does this matter now?

Voice is the one automation where a bad decision is audible to your customer. A per-minute rate that looks small becomes a real number once you count a busy month, and both of these platforms bill in a way that makes the headline figure incomplete - Vapi because the providers are separate, Retell because the bill is assembled from a dozen itemised lines. Getting the comparison right means you can commit to the platform your team can actually run, rather than discovering three weeks in that nobody in the building can debug a latency problem across four vendors.

Internal path

Where to go next from this page

These links are part of the A8gent learning and conversion path. Use them to move from concept, to diagnosis, to workflow build, to course.

Start with readiness

What you should be able to do after this

  • Price both on your real call minutes
  • Understand bundled vs itemised billing
  • Know which needs a developer
  • Decide without a sales call

How do you do it, step by step?

1. Count your real minutes before comparing rates

Take the calls you already take or miss in a month and multiply by average call length. The reference workload used across our platform reviews is 500 calls a month at three minutes each, so 1,500 billable minutes, which is roughly what a busy trades business, clinic or salon generates. Every rate on this page becomes a real number once you have that figure, and none of them mean anything before.

2. Price Vapi as four bills, not one

Vapi's $0.05 a minute as of August 2026 is its orchestration slice only. On top sit transcription, the language model, speech synthesis and telephony, each billed by its own provider. If you bring your own API keys, Vapi's share of those is zero and you pay the providers directly. Budget all four lines before comparing, because the $0.05 figure is genuinely the smallest part of the total.

3. Price Retell from its published component table

Retell publishes voice infrastructure at $0.055 a minute, text-to-speech at $0.015 and US Twilio telephony at $0.015 as of August 2026 - $0.085 you pay regardless. The language model swings from $0.003 to $0.16 and is the single biggest variable in the bill. Pick the model deliberately rather than accepting a default, because a careless choice quadruples what you pay.

4. Decide who owns a latency problem at 9am on a Monday

This is the question that actually separates them. On Vapi, a sluggish agent means working out which of four providers is slow. On Retell the pipeline is managed, so you have fewer levers and fewer places to look. Neither is better in the abstract - it depends entirely on whether the person who will get that call enjoys the debugging or dreads it.

5. Model your peak hour, not your monthly total

Concurrency is simultaneous calls, and it is the limit most businesses hit first. Vapi's Build plan includes 10 concurrent lines with extras at $10 per line per month. Retell includes 20 concurrent calls free, with extras at $8 per call per month above that, as of August 2026. A business taking 500 calls a month can easily see six at once on a Monday morning, so size on the rush.

6. Check the compliance line before you build anything

If you handle health data, price this first. Vapi lists HIPAA as a $2,000 per month add-on and Zero Data Retention at $1,000 per month on either plan. Retell prices PII removal at $0.01 a minute as a per-minute feature instead. Those are very different structures, and for a small clinic the difference can decide whether the project makes sense at all.

7. Run the same twenty real calls through both

Both are pay-as-you-go with no minimum, and Retell includes $10 in free credits to start. Build the same simple agent on each, run twenty genuine calls from real phones on real networks, and listen to the recordings with someone who knows what a good call sounds like in your business. That afternoon tells you more than any comparison page, including this one.

Pick Vapi when, pick Retell AI when

Pick Vapi when you have a developer who will still be there in six months, when you want your model, voice and telephony under your own accounts, and when cost visibility per component matters to you. Its $0.05 a minute orchestration fee as of August 2026 is the lowest platform slice in this pair, and bringing your own provider keys drops Vapi's share of model and voice cost to zero. It is also the better position on lock-in: leaving means replacing an orchestration layer rather than rebuilding a stack, because the providers underneath were always yours.

Pick Retell AI when you want to know what the bill will be before you build, and when the person configuring the agent is technically confident but not a developer. Retell publishes the most granular pricing in the category - voice infrastructure at $0.055 a minute, text-to-speech at $0.015, the language model at $0.003 to $0.16, US Twilio telephony at $0.015 - which means you can build an accurate forecast from the published page alone. That is genuinely rare here, and it pairs with a visual builder that gets a non-developer further than Vapi's API does.

Pick Retell also when concurrency is your worry rather than your rate. It includes 20 concurrent calls free against Vapi's 10, with extras at $8 per call per month rather than $10 per line, as of August 2026. For a business with a spiky Monday morning that difference matters more than a cent or two per minute, because a caller who hits a busy signal generally does not call back.

On call quality itself we are not going to separate them, because we have not run controlled latency or transcription tests and neither has published one we would trust. Both are serious platforms built by teams that understand real-time voice. The honest way to decide on quality is twenty real calls from real phones on each, listened to by someone who knows what a good call sounds like in your business.

The model choice matters more than the platform choice

The single most useful fact in this comparison is that Retell's published range spans $0.07 to $0.31 a minute, and that spread is wider than the gap between the two platforms. Voice infrastructure, text-to-speech and US telephony come to $0.085 a minute that you pay regardless. The language model swings from $0.003 to $0.16 - more than the entire rest of the stack combined at the top end.

The same logic applies on Vapi, except the model bills to your own provider account. On the reference workload of 1,500 minutes a month, Vapi's own charge is $75 and the all-in figure typically lands between $120 and $200, with the spread driven almost entirely by which voice and model you pick. So on both platforms, the highest-leverage decision you make is not which vendor to sign with.

The practical consequence is a testing order. Pick the cheapest model that passes a real test with your own calls, and only escalate where it visibly fails. A business booking haircuts or confirming appointments almost never needs a flagship model, and running one is the most common way to turn a $105 bill into a $465 one on identical work.

Watch the optional per-minute features too. Retell prices knowledge base, advanced denoising and safety guardrails at $0.005 a minute each, PII removal at $0.01, and AI quality assurance at $0.10 with the first 100 minutes free. That last one nearly doubles the cost of a cheap configuration if it is left running on everything. Use the free hundred minutes, then sample rather than monitor everything.

Where both of these are the wrong answer

If nobody in your business writes code, both are harder than they need to be. Vapi is explicitly a developer product - it hands you components and expects you to assemble them. Retell is more approachable and its visual builder covers real ground, but deep customisation still pushes you toward the API, and integrating it with your systems is technical work. A fully bundled provider costs more per minute and vastly less in total when the alternative is a half-finished integration nobody owns.

If you need one predictable number on a budget line, neither is comfortable. Vapi's total is assembled from four providers and Retell's from a dozen itemised lines, both of which move with usage. That transparency is a genuine strength for someone who wants to optimise and a genuine irritation for someone who just wants to know what next month costs. A bundled single-rate platform is the honest answer to that requirement.

And if your calls genuinely need human judgement rather than information handling, no voice agent on this page is the right answer at any price. Voice AI is very good at booking, qualifying, confirming and answering documented questions. It is poor at discretion, at reading a distressed caller, and at anything where being wrong is expensive. Deciding which of your calls fall into which category is worth more than choosing between these two.

At a glance

Vapi vs Retell AI decision table, published pricing as of August 2026

Decision factorVapiRetell AI
Billing structure$0.05/min orchestration, providers billed separately or via your own keysItemised per minute: $0.055 infrastructure, $0.015 TTS, $0.003-$0.16 model, $0.015 US telephony
All-in at 1,500 min/mo$75 to Vapi, typically $120-$200 with providers and telephony$105-$465 across the published $0.07-$0.31 range
Who can run itDeveloper required; components are yours to assembleVisual builder plus API; a technical operator gets further unaided
Concurrency included10 concurrent lines on Build, extras $10 per line/mo20 concurrent calls free, extras $8 per call/mo
CommitmentPay-as-you-go on Build; Scale is an annual contractPay-as-you-go, no minimums or contracts, $10 free credits
Compliance pricingHIPAA $2,000/mo add-on, Zero Data Retention $1,000/moPII removal $0.01/min; guardrails $0.005/min
Model choiceAny provider you like, swappable per componentSelectable from a range spanning $0.003 to $0.16/min
Switching costLow - providers are your own accounts, only orchestration is Vapi'sModerate - prompts and numbers port, configuration and history do not
Call qualityNot separated here - we ran no controlled tests; judge on twenty real callsNot separated here - we ran no controlled tests; judge on twenty real calls

What mistakes should you avoid?

  • Comparing Vapi's $0.05 orchestration fee against Retell's all-in range as though they measure the same thing
  • Accepting a default language model on Retell and paying flagship rates for booking appointments
  • Choosing Vapi without a developer who will still be around in six months
  • Sizing concurrency on monthly call volume instead of the busiest hour
  • Discovering the HIPAA or PII cost after the agent is built rather than before
  • Testing with your own voice in a quiet office instead of real callers on real phones
  • Leaving Retell's optional per-minute features switched on and never auditing the bill

FAQ

Which is cheaper, Vapi or Retell AI?
It depends on your model choice more than on the platform. On 1,500 minutes a month as of August 2026, Vapi is $75 of its own fee plus provider costs, typically landing around $120 to $200 all-in. Retell's published $0.07 to $0.31 range puts the same volume between $105 and $465. A well-tuned Retell configuration undercuts a careless Vapi stack, and the reverse is equally true.
Do I need a developer for either of these?
For Vapi, yes - it is explicitly a developer product and assembling a reliable agent takes real engineering. Retell pairs its API with a visual builder, so a technically confident operator can get further unaided, but deep customisation still pushes you toward the API. If nobody in the business writes code at all, look at a fully bundled provider instead.
How much of the bill is the language model?
On Retell, most of the variable part. Voice infrastructure, text-to-speech and US telephony come to $0.085 a minute regardless, while the model ranges from $0.003 to $0.16 as of August 2026 - more than the rest of the stack combined at the top end. On Vapi the same logic applies, except the model bills to your own provider account rather than through the platform.
Which one locks me in more?
Vapi slightly less, because your model, voice and telephony providers are your own accounts and Vapi supplies orchestration only - leaving means replacing that layer rather than the stack. Retell's switching cost is moderate: prompts and call flows are portable text and numbers can generally be ported, but the agent configuration and call history do not travel. Export transcripts from either regularly.
Is 10 or 20 concurrent calls enough?
For most small businesses, comfortably. Retell includes 20 free and Vapi 10 on the Build plan as of August 2026, with extras at $8 and $10 per line per month respectively. What matters is your peak hour rather than your monthly volume, because a caller who hits a busy signal generally does not call back. Count the busiest morning you had last quarter.
Can I move from one to the other later?
Yes, with real but manageable effort. Prompts and call scripts are plain text and port directly, and phone numbers can generally be moved between providers. What you rebuild is the agent configuration, the tool wiring and any platform-specific behaviour tuning. Keeping transcripts is the single most useful thing you can do, because they become the test set for whatever you move to.

Sources & further reading

Done for you

Want this built for your business?

Tell us the workflow. We scope, build, and ship the agent with guardrails and the numbers to prove it worked. The scoping call is free.

Free scoping call. You own the code.

Request a scoping call

Was this page helpful?