Skip to content
AI Receptionist Report

Head-to-head

Retell vs Vapi

The two leading developer platforms for voice agents, compared on published per-minute pricing and build philosophy for teams building their own receptionist.

By Network Editorial Team

Last updated

Disclosure: This website may receive compensation from companies mentioned on this page. The publisher may also have an ownership or financial relationship with certain featured providers. These relationships do not change our stated evaluation methodology. The publisher of this website has a financial interest in Torklio. Read the full disclosure.

RT

Retell AI

Developers building a custom voice agent

Developer voice-agent platform, pay-as-you-go from about $0.07 to $0.31 per minute.

Starting price: $0/mo

Read the Retell AI review
VP

Vapi

Developers who want maximum customization

Developer voice AI platform: $0.05/min platform fee plus pass-through model and telephony costs.

Starting price: $0/mo

Read the Vapi review

Decision summary

Both are platforms, not products: you build the receptionist. Retell offers a more opinionated build path with the conversation engine, telephony and APIs packaged together and a published all-in range of $0.07 to $0.31 a minute. Vapi leans toward maximum configurability, charging a $0.05 per-minute platform fee and passing speech, model and telephony costs through at cost, or $0 with your own API keys. Neither has a monthly fee, and both prices were verified on September 7, 2026. Choose Retell if you want to ship faster with sane defaults and a knowable per-minute cost; choose Vapi if you have strong opinions about every component or already hold the API keys. The rest of this page is for the engineer or agency doing the choosing, and for the owner deciding whether to hire one.

Feature differences

Customization

Retell AI
Deep, within an opinionated framework.
Vapi
Maximum; bring-your-own components.

Developer tooling

Retell AI
APIs and SDKs.
Vapi
APIs, SDKs, strong docs.

Scheduling and CRM

Retell AI
Custom function integration required.
Vapi
Custom tool integration required.

Human backup

Retell AI
None.
Vapi
None.

Billing

Retell AI
Per-minute component pricing with a published all-in range (verified).
Vapi
Platform per-minute fee plus pass-through costs at the provider's rates (verified).

Call recording and transcripts

Retell AI
Recorded as available (not independently verified).
Vapi
Recorded as available (not independently verified).

Multi-language

Retell AI
Recorded as supported (not independently verified).
Vapi
Recorded as supported (not independently verified).

Outbound calls

Retell AI
Yes (not independently verified).
Vapi
Yes (not independently verified).

Knowledge base

Retell AI
Built in: $0.005 per minute, 10 knowledge bases free, then $8 a month each (verified).
Vapi
Not priced on the pricing page.

Compliance add-ons

Retell AI
PII removal $0.01 per minute (verified).
Vapi
HIPAA add-on $2,000 a month; zero data retention $1,000 a month (verified).

Setup responsibility

Retell AI
Your developers or an agency build the receptionist behavior on top of the engine and telephony Retell packages.
Vapi
Your developers select and wire each model, voice and carrier, then build the behavior.

Ongoing maintenance

Retell AI
Prompt, function and telephony upkeep sit with your team; fewer swappable parts to re-test.
Vapi
Same upkeep plus the burden of tracking each third-party component's changes.

Pricing differences

Monthly fee and free usage

Retell AI
None; pay-as-you-go with $10 in free credits at signup.
Vapi
None on the Build plan; '60+ minutes included'.

Platform per-minute rate

Retell AI
$0.055 per minute for voice infrastructure.
Vapi
$0.05 per minute platform fee.

All-in per-minute cost

Retell AI
$0.07 to $0.31 published, adding LLM ($0.003 to $0.16), text-to-speech ($0.015 to $0.04) and telephony ($0.015).
Vapi
$0.05 plus speech-to-text, LLM and text-to-speech at cost ($0 with your own keys) plus telephony from your provider; no all-in figure published.

Concurrency

Retell AI
20 concurrent calls free; $8 per month per additional call.
Vapi
10 concurrent lines on Build; $10 per line per month beyond.

Phone numbers and extras

Retell AI
Numbers $2 per month each; knowledge base +$0.005/min, denoising +$0.005/min.
Vapi
Numbers and telephony billed by your carrier; SMS or chat $0.005 per message; HIPAA add-on $2,000 per month.

Use-case differences

Agency building receptionists for clients

Retell AI
Good fit; the published range makes client quotes easier.
Vapi
Good fit; bring-your-own keys can lower per-client cost.

Product team with model preferences

Retell AI
Adequate.
Vapi
Strong fit.

Business owner without engineers

Retell AI
Wrong category.
Vapi
Wrong category.

Software vendor serving a single trade, such as an HVAC dispatch tool

Retell AI
A packaged engine plus custom functions can reach a working agent for the trade's call types quickly; see the [HVAC](/industries/hvac) scenarios for what it must handle.
Vapi
Fine-grained control lets the vendor tune voice and model per market, at the cost of more integration work.

Healthcare product that needs HIPAA

Retell AI
PII removal is a $0.01 per-minute add-on; HIPAA terms not stated on the pricing page.
Vapi
HIPAA is a published $2,000 per-month add-on.

Multilingual customer base

Retell AI
Multi-language support recorded (unverified).
Vapi
Multi-language support recorded, with the added option of choosing the model per language (unverified).

Pros and cons

Retell AI

Pros

  • Faster path to a working agent
  • Published all-in range of $0.07–$0.31 a minute
  • 20 concurrent calls and 10 knowledge bases free
  • Packaged telephony at $2 per number

Cons

  • Engineering required
  • Effective rate varies four-fold with model and voice choices

Vapi

Pros

  • Deepest configurability
  • $0.05 platform fee; $0 pass-through with your own keys
  • Strong tooling

Cons

  • Engineering required
  • No published all-in figure
  • HIPAA costs $2,000 a month
  • No out-of-the-box workflow

Which one should you choose?

Choose Retell AI if…

  • You want to ship a working agent quickly.
  • You are fine with the platform's default components.
  • You want a per-minute range you can quote before you build.
  • Your team is small and would rather maintain fewer moving parts.

Choose Vapi if…

  • You want to pick every model, voice and carrier.
  • You already hold API keys or contracts with a model or telephony vendor and want the pass-through cost to be $0.
  • You are building a product where voice quality tuning is a differentiator.

Side by side

RT
Retell AI

Developers building a custom voice agent

Starting price
$0/mo
24/7 answering
Yes†
Appointment scheduling
Partial†
Call transfers
Yes†
Human backup
No
CRM integration
Partial†
Developer access (API)
Yes
Industry specialization
No†
Test score
Sample only
Review status
Editorial Preview
VP
Vapi

Developers who want maximum customization

Starting price
$0/mo
24/7 answering
Yes†
Appointment scheduling
Partial†
Call transfers
Yes†
Human backup
No
CRM integration
Partial†
Developer access (API)
Yes
Industry specialization
No†
Test score
Sample only
Review status
Editorial Preview

† Not independently verified: based on the provider's public materials or editorial knowledge, not yet checked against a dated source. “Verify” means no data has been entered yet. Test scores appear only when the 10-call challenge has been completed for that provider.

How we compared them

Retell and Vapi are recorded in our database as developer platforms, and we compare them as such rather than as finished receptionists. Pricing on both sides was checked on September 7, 2026. Retell publishes its component rates and sums them to an all-in range; Vapi publishes its platform fee and states that models and telephony are passed through at cost. The feature rows come from each company's public developer materials, and we have not built an agent on either platform for the lab. The edges in the table therefore describe pricing transparency and design philosophy, not measured latency, accuracy or uptime.

The workflow after the call

On a developer platform the workflow after the call is whatever you write. Both Retell and Vapi expose recordings and transcripts (recorded as available, unverified) and both let the agent call out to your own functions or tools during the conversation, which is how scheduling and CRM updates get built. Retell prices a built-in knowledge base at $0.005 a minute; on Vapi, knowledge retrieval is something you wire in. Neither ships a qualification model, a customer history, a calendar conflict check or a screened-transfer policy; if you want those, you design them, host them and maintain them. That is the honest reason a business owner should not be on this page unless they have a developer.

Total cost of ownership considerations

Both platforms bill by the minute with no monthly fee, and in both cases the platform figure is only one layer of the effective rate. Retell's $0.055 voice-infrastructure fee becomes $0.07 to $0.31 a minute once a language model, a voice and a carrier are added, so 1,000 minutes costs between about $70 and $310 depending on the components chosen. Vapi's $0.05 platform fee is the floor; with third-party speech and model services billed at cost the total lands somewhere above it, and with your own API keys the pass-through drops to $0 and you pay those vendors directly. Concurrency is the other line: Retell includes 20 simultaneous calls and charges $8 a month for each beyond, Vapi includes 10 lines and charges $10 a month each. Compliance can dominate: Vapi's HIPAA add-on is $2,000 a month. The largest cost on either platform is not metered at all: the engineering time to build, test, monitor and update the agent as models and APIs change.

What would change this verdict

  • A measured per-minute cost for the same reference build on each platform would tell us where in Retell's range a realistic receptionist lands and what Vapi's pass-through actually adds.
  • A lab build of the same receptionist on each platform, scored on the ten-call challenge, would replace design-philosophy edges with measured ones.
  • Either vendor shipping a packaged receptionist template with native scheduling would move it toward the turnkey category and out of this comparison.
  • Documented uptime or latency guarantees from either would become a decisive row for agencies with service-level commitments.

Frequently asked questions

Can a small business use Retell or Vapi directly?

Not without a developer. Choose a packaged receptionist instead.

Which is cheaper per minute?

Vapi's platform fee is lower ($0.05 against Retell's $0.055), but Vapi's total depends on the models and carrier you choose. Retell publishes an all-in range of $0.07 to $0.31 a minute (both verified September 7, 2026). With your own API keys, Vapi's pass-through is $0 and you pay those vendors directly.

Which is better for a no-code team?

Neither. Synthflow's enterprise contracts (from $30,000 a year) include implementation; for a small business, a packaged product such as Torklio, Goodcall, Rosie, Dialzara or Frontdesk is the realistic answer.

Do Retell or Vapi provide human receptionists?

No. Any human escalation is something your developers build, typically as a transfer to your own staff. For a managed human layer, see Smith.ai.

If I want a finished receptionist for a trade, where should I look instead?

Packaged products such as Torklio, Goodcall or Rosie. AI Receptionist Report's publisher holds a financial interest in Torklio.

Sources

  1. Retell AI pricing (accessed 2026-09-07)
  2. Vapi pricing (accessed 2026-09-07)