Retell AI vs Vapi: Pricing, Features and Which to Pick
Two of the most searched voice agent platforms, compared line by line on published prices and features.
By , research by the BestBots team · Updated Oct 18, 2026
Fact checked against each company's published pricing, documentation and verified user ratings. How we score
Retell AI is the better pick for most teams: it includes 20 concurrent calls and HIPAA at no extra cost and has a visual builder. Vapi wins for developer teams that want to choose every provider and squeeze the per minute rate, since its lowest published stack is about $0.082 per minute against Retell's $0.095.
Side by side
| Spec | Retell AI | Vapi |
|---|---|---|
| BestBots Score | 4.6 | 4.3 |
| Starting price | $0.055 per minute | $0.05 per minute |
| Pricing model | per minute, pay as you go | per minute, pay as you go plus optional monthly packages |
| Free trial | $10 in free credits | $5 in free credits |
| Per minute cost | $0.055 voice infra plus LLM ($0.0016 to $0.32), voice ($0.015 to $0.10) and telephony ($0.015 US Twilio or Telnyx) | $0.05 hosting plus STT ($0.0095 to $0.0099), LLM ($0.0077 to $0.0452) and voice ($0.0146 to $0.0238) |
| Latency | About 600 ms, per Retell's homepage | Under 500 ms average, per Vapi's homepage |
| Telephony | Retell numbers via Twilio, Vonage, or your own SIP trunk at no Retell charge | Free Vapi numbers, SIP and WebRTC, or Twilio ($0.008 inbound, $0.014 outbound), Vonage ($0.00814) or Telnyx ($0.0055) per minute |
| Concurrency | 20 concurrent calls free, then $8 per line per month | 4 free, 10 on Core, 30 on Pro, extra lines $10 per month |
| Languages | Multilingual agents supported; the dashboard lists current languages | Depends on providers: Deepgram 100+, Google STT 125+, Azure TTS 140+ languages |
| HIPAA | Yes, listed as HIPAA ready | Add on at $2,000 per month with a signed BAA; cannot be combined with zero data retention |
| SOC 2 | SOC 2 Type II | Listed as SOC 2, HIPAA and PCI compliant for enterprise |
The short answer
Both platforms do the same job: they orchestrate speech recognition, an LLM and a synthetic voice on a phone call, and both charge a platform fee per minute with the models billed on top. The difference is who they are built for.
Retell is a product. It has a drag and drop canvas, testing and QA tools, a prompt based agent builder and generous defaults (20 concurrent calls, $10 in credits, HIPAA). A marketer or ops lead can launch an agent, and a developer can extend it with the API.
Vapi is a developer platform. It describes itself as API native, publishes a hosting fee and passes models through at cost, and lets you swap transcription, LLM and voice vendors freely. It is cheaper per minute at the bottom of the range, but you pay for concurrency, support and compliance as you grow.
Pricing
| Line item | Retell AI | Vapi |
|---|---|---|
| Platform fee | $0.055 per minute (includes speech to text) | $0.05 per minute hosting |
| Speech to text | Included | $0.0095 to $0.0099 per minute (Deepgram) |
| LLM | $0.0032 to $0.16 per minute | $0.0077 to $0.0452 per minute (OpenAI) |
| Voice | $0.015 to $0.10 per minute | $0.0146 to $0.0238 per minute (ElevenLabs) |
| Telephony | $0.015 per minute US Twilio, custom SIP free | Vapi telephony free, Twilio $0.008 in or $0.014 out |
| Free credits | $10 | $5 |
| Concurrency included | 20 | 4 (10 on $29 Core, 30 on $999 Pro) |
| Extra concurrency | $8 per line per month | $10 per line per month |
| HIPAA | Included | $2,000 per month |
Prices are as listed on each vendor's pricing page in October 2026.
Budget stack. Retell with Gemini 3.5 Flash Lite ($0.0096), a Retell voice ($0.015) and US Twilio ($0.015) totals $0.0946 per minute. Vapi at the low end of every range on Vapi telephony totals $0.0818 per minute. At 5,000 minutes that is $473 versus $409. At 20,000 minutes, $1,892 versus $1,636.
Premium stack. Retell with Claude 5 Sonnet ($0.064) and ElevenLabs Flash ($0.040) totals $0.174 per minute. Vapi at the top of every range plus Twilio outbound totals $0.1429. Note that Retell lists more expensive model options (GPT 5.5 at $0.16 per minute, ElevenLabs v3 at $0.10), so its ceiling is higher if you choose them.
The concurrency and compliance swing. Say you run a clinic line with 3,000 minutes a month, need 10 simultaneous calls and need HIPAA. On Retell's budget stack: 3,000 x $0.0946 = $283.80, with concurrency and HIPAA included. On Vapi: 3,000 x $0.0818 = $245.40, plus $29 for Core to reach 10 lines, plus $2,000 for HIPAA, totals $2,274.40. The per minute gap is small; the add ons decide it.
For more volumes, use the cost calculator or read the cost per minute breakdown.
Features
| Feature | Retell AI | Vapi |
|---|---|---|
| Builder | Drag and drop canvas plus prompt agents | API and dashboard configuration |
| Latency claim | About 600 ms | Under 500 ms average |
| Model choice | Retell's listed LLMs and voices, or custom LLM | OpenAI, Anthropic, Gemini, Groq and more, pass through |
| Telephony | Twilio, Vonage, SIP | Vapi numbers, SIP, Twilio, WebRTC |
| Languages | Multilingual, dashboard lists current options | Provider dependent: Deepgram 100+, Google STT 125+, Azure TTS 140+ |
| Compliance | SOC 2 Type II, HIPAA, GDPR, ISO 27001 | SOC 2, HIPAA (add on), PCI listed for enterprise |
| Integrations | HubSpot, Cal.com, Make, n8n, GoHighLevel | LLM, voice and speech vendors, tools via API |
| Free tier support | Standard support | Discord community |
| Data retention (free) | Not published | 14 days |
Retell's edge is completeness out of the box: testing tools, a visual flow, CRM and calendar connectors. Vapi's edge is choice: it lists more providers at each layer, and a developer can wire up any LLM or voice that has an API.
Who wins for which buyer
A small business with no developer. Retell. You can build in the canvas, start with $10 in credits and stay under the 20 call concurrency limit without paying for a package.
A startup building a voice product. Vapi. Pass through pricing, provider choice and a hosting fee of $0.05 per minute give you control over margin. Budget $29 or $999 per month once you need more than 4 concurrent calls.
A healthcare practice or any HIPAA workload. Retell. HIPAA is included; on Vapi it adds $2,000 per month.
An outbound campaign with high concurrency. Close call. Retell includes 20 lines and charges $8 per extra line plus $0.005 per batch dial; Vapi charges $10 per extra line. At 50 lines, Retell's extra 30 cost $240 per month and Vapi's extra 46 on the free tier cost $460, or $400 on Core. Retell is cheaper on lines; Vapi is cheaper per minute.
Reviews and trust
Retell holds a 4.9 Trustpilot score from 810 reviews, with praise for setup speed and voice quality and some complaints about latency pauses and support replies. Vapi's Trustpilot profile is unclaimed with a 2.4 score from 15 reviews citing pauses, documentation gaps and charges beyond the platform fee. Fifteen reviews is a small sample, and Vapi reports 1M+ developers, but the gap is worth noting.
Verdict
Pick Retell AI if you want a platform the whole team can use, need HIPAA, or want to stay on the free concurrency tier. Pick Vapi if engineers own the agent, you want to choose every provider, and you will run enough volume that saving about a cent per minute outweighs paying for lines and compliance. For a third option with the lowest all in rate, see Telnyx in our best AI voice agents guide.
Frequently asked questions
Is Retell AI cheaper than Vapi?
Not per minute. On the cheapest published components Vapi comes to about $0.0818 per minute on free Vapi telephony, while Retell comes to about $0.0946 with US Twilio. Retell gets cheaper once you count concurrency and compliance: it includes 20 concurrent calls and HIPAA, where Vapi needs a $29 or $999 package for more lines and $2,000 per month for HIPAA.
Which is easier to use, Retell AI or Vapi?
Retell. It has a drag and drop flow canvas, built in testing and a quick start that gets a phone agent live in about 15 minutes. Vapi describes itself as API native and is built for developers who are comfortable configuring providers in code.
Does Vapi include the LLM in its price?
No. Vapi's $0.05 per minute is a hosting fee. Transcription, the LLM and the voice pass through at cost, listed as $0.0095 to $0.0099, $0.0077 to $0.0452 and $0.0146 to $0.0238 per minute respectively.
Does Retell AI include the LLM in its price?
No. Retell's $0.055 voice infra rate includes speech to text only. The LLM is $0.0032 to $0.16 per minute depending on the model, the voice is $0.015 to $0.10 per minute, and US Twilio telephony is $0.015 per minute.
Which has lower latency?
On vendor claims, Vapi lists under 500 ms average and Retell about 600 ms. Neither publishes a shared test method, and real latency depends heavily on the LLM and voice you pick, so measure both on your own script.
Can I bring my own phone numbers?
Yes on both. Retell supports Twilio, Vonage and SIP trunking and does not charge for custom telephony. Vapi's own telephony and SIP transport are free, and Twilio adds $0.008 inbound or $0.014 outbound per minute.
Which is better for healthcare?
Retell, for most clinics. It lists HIPAA with no add on fee plus SOC 2 Type II. Vapi offers HIPAA, but the $2,000 per month add on only pays off at very high volume.