Skip to content

Vapi vs Retell for Agencies: Pricing, Analytics and Client Fit (2026)

Vapi vs Retell for agencies reselling voice agents: per-minute costs, concurrency, call analysis, webhooks and client reporting, as of November 2026.

By , Founder & CEO

Updated · Published · 9 min read

Two voice waveforms side by side comparing Vapi and Retell for agencies

Short answer

Pick Vapi if you want to assemble your own stack (transcriber, model, voice, telephony) and pay models at provider cost on top of a $0.05/min hosting fee. Pick Retell if you want bundled per-minute pricing ($0.07 to $0.31/min), 20 free concurrent calls and built-in sentiment and success fields. Both work for agencies; the difference is control versus packaging.

Key takeaways

  • Vapi charges $0.05/min hosting plus model, voice and telephony costs at provider rates; Retell quotes $0.07 to $0.31/min with components priced on one page.
  • Retell includes 20 concurrent calls free; Vapi includes 4 on pay-as-you-go and 10 on the $29/month Core plan.
  • Retell gives call summary, user sentiment, call success and voicemail detection by default. Vapi gives summary, structured data and a success evaluation with eight rubrics.
  • Vapi plans include a set number of organizations (2, 2, 10 or custom), which matters if you separate clients by org.
  • Whichever you pick, budget for per-client reporting and billing: neither pricing page lists a white-label client portal.

Agencies that resell AI receptionists and outbound callers usually end up choosing between Vapi and Retell AI. Both let you build phone and web voice agents, both expose call data over an API and webhooks, and both bill by the minute. The differences show up in how the minute is priced, how many calls can run at once, what post-call analysis you get out of the box, and how much work it takes to turn call data into something a client will read.

Everything below comes from the official Vapi and Retell pricing pages and documentation, checked as of November 2026. Prices change often in this market, so confirm on the live pages before you quote a client.

What is the short answer to Vapi vs Retell?

Vapi is a developer platform where you pick each layer of the call: transcriber, language model, voice and telephony. You pay Vapi a hosting fee and pay the model providers at their own rates. Retell sells a more packaged product: one per-minute price built from voice infrastructure, a chosen LLM, a chosen voice and telephony, with post-call analysis fields already defined.

If your team likes to tune every component and you have clients with unusual requirements, Vapi tends to fit. If you want predictable per-minute quotes and more concurrency before paying extra, Retell tends to fit. Many agencies run both, which is why reporting across platforms matters (more on that below).

How do Vapi and Retell compare side by side?

Vapi vs Retell, from official pricing pages and docs, as of November 2026
FactorVapiRetell AI
Pricing model$0.05/min Vapi hosting, plus transcriber, model, voice and telephony at provider rates$0.07 to $0.31/min, built from voice infra, LLM, voice and telephony line items
Free credits$5$10
Paid plan tiersNo Success Package ($0), Core ($29/mo), Pro (10% of hosting fee, $999/mo minimum), Premier (custom)Pay-as-you-go, Enterprise (custom)
Included concurrent calls4 (pay-as-you-go), 10 (Core), 30 (Pro), custom (Premier)20 free
Extra concurrency$10 per line per month$8 per concurrent call per month, or burst at +$0.10/min
Organizations2, 2, 10 or custom by planNot listed on the pricing page
Raw data retention14, 30 or 180 days by plan; custom on PremierNot listed on the pricing page
Default post-call analysisSummary, structured data, success evaluation (8 rubrics)Call summary, user sentiment, call successful, in voicemail
Call-complete webhookend-of-call-report server messagecall_ended, then call_analyzed when analysis finishes
HIPAA$2,000/month add-on with BAANot listed on the pricing page; ask sales
Agency or white-label programNot listed on the pricing pageNot listed; a Solution Partner Program is mentioned without terms

How does Vapi pricing work for an agency?

Vapi's pricing page lists a $0.05 per minute hosting fee on the pay-as-you-go package, with models "passed through at cost." On top of that you pay for each component. The page shows these example ranges:

  • Transcriber (Deepgram): $0.0095 to $0.0099 per minute.
  • Language model (OpenAI): $0.0077 to $0.0452 per minute.
  • Voice (ElevenLabs): $0.0146 to $0.0238 per minute.
  • Telephony: Vapi Telephony/SIP, Vapi WebSockets and Daily WebRTC are listed as free. Twilio is $0.008/min inbound and $0.014/min outbound, Vonage $0.00814/min and Telnyx $0.0055/min, charged by the provider.

The Core plan is $29 per month and raises concurrency to 10, phone numbers to 5 and retention to 30 days. Pro costs 10% of your hosting fee with a $999 monthly minimum, and adds 10 organizations, 30 concurrent calls, 180-day retention and role-based access control. Premier is custom and adds SSO and dedicated SIP.

For agencies, two Vapi details matter more than the headline rate. First, organizations: if you want each client in its own org for clean separation of numbers, keys and billing, the lower plans include only two. Second, retention: on pay-as-you-go, raw call data is kept for 14 days. If you promise clients a 90-day call history, you need to copy calls out of Vapi before they age out.

How does Retell pricing work for an agency?

Retell's pricing page quotes $0.07 to $0.31 per minute for voice agents and starts you with $10 in credits. The default calculator example comes to $0.11 per minute: $0.055 Retell voice infrastructure, $0.04 LLM and $0.015 text-to-speech. Telephony is listed at $0.015 per minute for the default country, varies by country, and has no charge for SIP trunking or bring-your-own telephony.

Voices and models change the rate. ElevenLabs voices range from $0.04 (Flash) to $0.10 per minute, while Retell's platform voices and several others are $0.015. LLM rates on the page range from $0.008 (Gemini 3.1 Flash Lite) to $0.08 (Claude 4.6 Sonnet) on the standard tier, with higher fast-tier rates for most models.

Retell also lists billing rules you should pass through to your own quotes:

  • Calls under 10 seconds with a dynamic greeting bill as 10 seconds.
  • Prompts averaging over 4,000 tokens increase billed time proportionally.
  • Billing continues during silence and hold.
  • After a transfer, the AI fee stops but telephony continues.
  • Add-ons are priced separately, for example Knowledge Base at +$0.005/min, PII removal at +$0.01/min and branded outbound calls at +$0.10 per call.

What does a month of calls cost on each? (worked example)

This is a worked example, not a benchmark. It uses the per-minute rates on each pricing page as of November 2026 for one client with 3,000 inbound minutes a month. Your real cost depends on the model, voice, prompt length and call mix.

Worked example: 3,000 inbound minutes per month
SetupPer-minute build-upApprox. per minuteApprox. per month
Vapi, lower-cost components, Twilio inbound$0.05 + $0.0095 + $0.0077 + $0.0146 + $0.008$0.090$269
Vapi, higher-cost components, Twilio inbound$0.05 + $0.0099 + $0.0452 + $0.0238 + $0.008$0.137$411
Retell, default calculator setup, Retell telephony$0.055 + $0.04 + $0.015 + $0.015$0.125$375

The ranges overlap, so headline price rarely decides it. What changes the bill more is concurrency (a busy receptionist line needs headroom), add-ons such as HIPAA or PII removal, and whether you mark up minutes or pass them through at cost. Whatever you choose, record the platform cost per call so you can see margin per client later.

Which gives better post-call analytics for client reports?

Clients rarely care about tokens. They care whether calls were answered, whether the agent booked the appointment and whether callers were happy. The two platforms approach this differently.

Retell: fixed fields out of the box

Retell's post-call analysis adds four built-in fields to every call: call_summary, user_sentiment, call_successful (against success criteria you define) and in_voicemail. You can add custom extraction fields on top. Retell notes that sentiment is inferred from the transcript by an LLM, not from tone of voice. One practical detail: analysis results are not in the call_ended webhook, so you need to listen for call_analyzed or fetch the call afterwards.

Vapi: configurable analysis

Vapi's call analysis produces a summary, structured data extracted to a JSON schema you define, and a success evaluation. The success evaluation can use eight rubrics, from pass/fail to a 1 to 10 numeric scale or a checklist. The docs now recommend structured outputs over the older analysisPlan and point to scorecards for success evaluation. There is no fixed sentiment field in that list, so if a client wants sentiment you define it yourself as a structured output.

In short: Retell gives you a consistent schema across every agent with no setup, which is easier when reporting across many clients. Vapi gives you more freedom, which is better when each client measures success differently, but you have to keep the definitions consistent yourself.

How do webhooks and call data differ?

Vapi sends server messages to a server URL you configure. The one that matters for reporting is end-of-call-report, which carries endedReason, the call object and an artifact with the recording, a transcript string and a message array. Most Vapi messages are informational and need no response.

Retell sends agent webhooks for call_started, call_ended, call_analyzed and several transfer events. Each request carries an x-retell-signature header that you verify with your API key against the raw request body, and Retell publishes an IP address you can allowlist.

Both also offer list-calls APIs, which is how a reporting tool backfills history and catches any webhook that was missed. If you connect either platform to GetFlowBiz, the Vapi integration and Retell integration use the API key to sync assistants or agents and their calls, and accept the webhook for near real-time updates.

How should an agency choose between Vapi and Retell?

  1. List your client's must-haves. Concurrency at peak hour, languages, HIPAA, call transfer, outbound campaigns and data retention.
  2. Price one real call flow on both. Use your actual prompt length, model and voice, not the cheapest line on the page. Include telephony and add-ons.
  3. Check concurrency headroom. A clinic with 8 lines at 9am needs more than Vapi's 4 pay-as-you-go lines; Retell's 20 free lines may be enough.
  4. Decide how you will separate clients. Separate organizations or workspaces per client keep keys, numbers and data apart. Check how many your plan includes.
  5. Define success per client. Write the success criteria (booked, qualified, transferred) before launch so the analysis fields mean something in reports.
  6. Plan reporting and billing. Decide where call data, cost per call and invoices will live, because neither pricing page lists a client portal for agencies.

What about client reporting and billing on top?

This is where agencies lose time. Vapi and Retell dashboards are built for the builder, not for your client. Clients want a monthly view of calls answered, calls missed, average call length, how many calls hit the goal, and what they owe you. Doing that by hand means exporting calls, building charts and sending invoices separately.

GetFlowBiz connects Vapi, Retell, Bland, ElevenLabs Agents, Synthflow and other voice platforms by API key, maps every call to the same fields, and shows answer rate, missed calls, average call length, sentiment and transcripts in a white-label client portal with invoices and online payments. It is free right now through the Founding Member program: the first 100 accounts stay free forever. You can sign up free or see all integrations.

Frequently asked questions

Is Vapi cheaper than Retell?

Not by default. As of November 2026, Vapi charges $0.05/min hosting plus model, voice and telephony at provider rates, while Retell quotes $0.07 to $0.31/min all-in. With low-cost components Vapi can come in under Retell; with premium models and voices it can cost more. Price your real call flow on both.

How many concurrent calls do Vapi and Retell include?

Retell includes 20 concurrent calls free and charges $8 per extra concurrent call per month, with an optional burst at +$0.10/min. Vapi includes 4 on pay-as-you-go, 10 on Core ($29/month) and 30 on Pro, and charges $10 per extra line per month.

Does Retell or Vapi provide caller sentiment?

Retell includes a user_sentiment field in its default post-call analysis, inferred from the transcript. Vapi's call analysis gives a summary, structured data and a success evaluation; to get sentiment on Vapi you define it as a structured output field.

Can an agency white-label Vapi or Retell for clients?

Neither pricing page lists a white-label or agency plan as of November 2026. Agencies usually give clients a separate branded portal or report that pulls call data from the Vapi or Retell API, so the client never logs into the voice platform.

Which platform is better for HIPAA clients?

Vapi lists HIPAA as a $2,000/month add-on with a Business Associate Agreement. Retell's public pricing page does not list HIPAA terms, so ask Retell sales directly. In both cases confirm the BAA scope with the vendor before taking on a healthcare client.

Sources

  1. Vapi pricing
  2. Retell AI pricing
  3. Vapi docs: Call analysis
  4. Vapi docs: Server events
  5. Retell docs: Post-call analysis
  6. Retell docs: Webhooks

Kishan Makani

Founder & CEO, GetFlowBiz

Kishan builds GetFlowBiz, the client portal and operations platform for automation and AI agencies. He writes about running, pricing and reporting on automation work. More from Kishan

Keep reading

Lock in GetFlowBiz free, forever

Founding Members keep every feature free for life. Only 27 of 100 spots are left.

73/100 founding spots claimed27 left
Vapi vs Retell for Agencies: Pricing and Analytics | GetFlowBiz