Vapi vs Retell
Vapi vs Retell: a fair comparison, pricing first.
Vapi and Retell AI both let you build voice agents that answer and place phone calls, and both charge by the minute. They differ in what that minute includes. Vapi charges $0.05 a minute for hosting and passes model costs through at cost. Retell charges $0.055 a minute for its voice infrastructure and prices the model, voice and telephony itself, for $0.07 to $0.31 a minute all in.
Below is a side-by-side comparison from each company’s own pricing pages and documentation as of Nov 2026, written to be fair to both. After it: how to choose, and where VocalLabs fits if you want to keep one of them or neither is quite right.
Sources: vapi.ai/pricing · docs.vapi.ai · retellai.com/pricing · docs.retellai.com · Nov 2026
- AI agent00:00
Hello, you have reached Harbour Pharmacy. Are you calling about a prescription?
- Caller00:04
Yes, I need to refill my blood pressure tablets.
- AI agent00:07
I can help with that. Could you confirm your date of birth?
- Caller00:10
Fourteenth of March, 1968.
- System00:12
Tool call · prescription lookup · refill eligible
- AI agent00:14
Thank you. Your refill will be ready to collect after three this afternoon.
- System00:19
Billed per minute · platform fee + model + voice + telephony
Vapi vs Retell AI, line by line.
Every figure below comes from the two companies’ own pages. Neither column is the winner; the right one depends on how you build and what you are willing to manage.
| Vapi | Retell AI | |
|---|---|---|
| Describes itself as | The developer platform for building voice AI agents | A platform to build, test, deploy and monitor AI voice and chat agents |
| Headline price | $0.05/min hosting, pre-paid, plus models and transport | $0.07–$0.31/min for voice agents, all in |
| Platform’s own per-minute fee | $0.05/min hosting | $0.055/min voice infrastructure |
| Models and voices | Passed through at cost with no markup, or your own provider keys | Priced per minute by Retell, e.g. GPT 5 nano $0.0016, Retell voices $0.015, ElevenLabs Flash $0.040 |
| Telephony | Vapi telephony and SIP free of transport fees; Twilio $0.008/min in, $0.014/min out, billed by Twilio | Retell numbers $2/month plus $0.015/min; no charge for SIP trunking or custom telephony |
| Free credits | $5 | $10 |
| Concurrency included | 4 on usage-based, 10 on Core, 30 on Pro; extra lines $10/month | 20 on pay as you go; extra calls $8/month each |
| Monthly packages | Core $29/month; Pro 10% of hosting, $999/month minimum; Premier custom | Enterprise, custom pricing |
| How you build | Dashboard, web SDKs and a CLI; assistants, squads and tests as files in git | Single-prompt or conversation-flow agents; dashboard, API, Node.js and Python SDKs, MCP |
| Billing detail | Pre-paid; no minimum minutes on usage-based | Per second, totalled per cycle; silence billed; failed connections not billed |
| HIPAA | Add-on, $2,000/month | HIPAA-ready with a BAA; SOC 2 certified |
From vapi.ai/pricing, docs.vapi.ai, retellai.com/pricing and docs.retellai.com as of Nov 2026, in US dollars. Both companies change prices and plans often; check the pages before you commit. · Source: Vapi pricing
Choose Vapi if… choose Retell if…
- 01Vapi
You want models at cost
Vapi passes model and speech costs through with no markup, or bills them to your own provider keys, so its own line is the hosting fee.
- 02Vapi
Your team works in code
Vapi keeps assistants, squads, tools and tests as files in git, reviewed through pull requests, with a CLI alongside the dashboard.
- 03Vapi
You need many assistants that hand off
Vapi’s squads pass a conversation between specialised assistants with its context preserved.
- 04Retell
You want one bill for the call
Retell runs telephony, speech and models together and prices each on one page, so the whole call is billed in one place unless you bring your own telephony.
- 05Retell
You need concurrency from day one
Retell includes 20 concurrent calls on pay as you go; Vapi’s usage-based plan starts at 4 without a package.
- 06Retell
Your calls follow fixed steps
Retell recommends conversation-flow agents, split into nodes, for calls that must follow a set order.
How to decide between Vapi and Retell.
- 01
Price one real minute
Add up the platform fee, the model, the voice and telephony for the setup you would actually run. Both pricing pages have estimators; use the same model and voice on each.
- 02
Count your peak concurrency
Simultaneous calls drive the add-on bill: Vapi charges $10 per extra line a month, Retell $8 per extra concurrent call.
- 03
Price compliance in early
If you handle health data, HIPAA is a $2,000-a-month add-on on Vapi, and Retell lists itself as HIPAA-ready with a BAA. Ask each for the terms in writing.
- 04
Run the same script on both
Both give free credits. Build the same agent on each, call it from a real phone, and compare latency, interruptions and transcripts.
Where VocalLabs fits.
VocalLabs does not have to replace either. It fits in three places: under the agent you already have, instead of it, or as a product you sell.
Questions buyers ask first.
Keep the agent you chose. Or build the next one here.
Tell us which platform you are on, or leaning towards, and what your calls look like: carrier, volume and channels. We will show you where VocalStack Voice or the VocalLabs agent builder would change anything, and where it would not.
