Retell AI vs Vapi (2026): Pricing, Model, and How They Bill
Two developer-first voice platforms, priced very differently. Retell assembles a per-minute rate from components; Vapi charges a low hosting floor plus pass-through provider costs.
Retell AI vs Vapi compared for 2026: pricing models, what the per-minute rate includes, free tiers, concurrency and compliance — plus where Flowyte fits for teams who want an all-in agent.
Free credits. No credit card. Zero balance pauses — never bills.
Retell AI vs Vapi
Retell AI and Vapi are both developer-first platforms for building AI voice agents, and both are usage-based — but they bill in almost opposite ways. Retell is composable: you assemble a per-minute rate from voice infrastructure, text-to-speech, the model and telephony, each billed separately. Vapi charges a low $0.05/min platform hosting fee and passes the model-provider costs (speech-to-text, model, text-to-speech) through on top, at cost or $0 if you bring your own API keys. Figures below are from each product’s live pricing page, as of July 2026.
| Dimension | Retell AI | Vapi |
|---|---|---|
| Pricing model | Usage-based, composable — assemble the rate from components | Platform hosting $0.05/min + pass-through provider costs |
| Per-minute rate | $0.07–$0.31/min assembled (infra $0.055 + voice + model + telephony) | $0.05/min hosting only; STT/LLM/TTS billed on top |
| What the rate includes | Voice infrastructure; your chosen TTS, model and telephony billed separately | Vapi’s hosting slice only; providers separate (or $0 with your own keys) |
| Bring your own model keys | Model-agnostic component menu (pick per call) | Yes — $0 provider cost on Vapi’s side with your own keys |
| Getting started | $0 to start, $10 free credits, 20 free concurrent calls | Usage-based Build plan with some included minutes |
| Compliance add-ons | See their pricing page | HIPAA $2,000/mo, Zero Data Retention $1,000/mo |
| Best for | Developers who want a composable, model-agnostic per-minute stack they tune component by component | Engineers who want the lowest hosting floor and to bring their own provider keys |
Figures from each product's live pricing page, as of July 2026 — Retell AI pricing, Vapi pricing. A dash means it wasn't published there.
Where Flowyte fits
Both Retell and Vapi are build-it-yourself platforms — you assemble the stack, pick or bring the providers, and predict a per-minute number that depends on the components you wire. Flowyte is API-first too, but ships the whole agent studio at one all-in $0.11/min that already includes the voice, the model and transcription. There is nothing to assemble, no keys to bring, and no pass-through invoices to reconcile — a 3-minute call is about $0.33, full stop.
Flowyte’s $0.11/min pay-as-you-go already includes voice, model and transcription — no infra + TTS + model + telephony stack to price out, and no pass-through provider bills. Numbers are $2/mo and transfers $0.03/min.
Describe your business and AI drafts the agent’s persona, goals, guardrails and Skills, which you shape in plain English and publish — with a pre-publish report, versioning and rollback. Everything the studio does is also available over the REST /api/v1 with scoped keys and webhooks.
Instead of building and hosting a tool server for live lookups, Flowyte connects your Postgres or MySQL with a guided form, scopes what the agent can see, and freezes every query you approve — order status on a call without writing code.
Connect your AI agent to your database
FAQ
Retell AI vs Vapi FAQ
Common questions about Retell AI, Vapi and Flowyte
Retell AI vs Vapi — what’s the main difference in pricing?
Which is cheaper, Retell AI or Vapi?
Do Retell AI and Vapi have free trials?
Where does Flowyte fit against Retell AI and Vapi?
Ready to build your agent?
Start free with Flowyte — describe your business, shape the agent in plain English, and go live on phone and chat. No contracts, no setup fees, voice from $0.11/min.