Pricing breakdown
PolyAI pricing: what is actually published
The short answer to the most common search here is that PolyAI does not publish pricing. There is no rate card, no self-service signup and no free tier; pricing is quoted per client through a sales process, billed on usage.
That is a legitimate enterprise model, and it is also the reason this page exists: if you are being asked to compare a quoted enterprise number against published per-minute rates, you need to know what to make comparable before you can judge it.
| Question | PolyAI | Platforms that publish |
|---|---|---|
| Published rate card | None. | Vapi $0.05/min platform fee; Retell $0.07 to $0.31/min all in. |
| Self-service signup | No. | Yes on both. |
| Free tier | No. | Trial credits are common. |
| Pricing model | Custom, usage-based, quoted per client. | Per minute, with optional committed volume. |
| Evaluation path | Sales-led, with scoping. | Sign up and test the same day. |
What to insist on in a quote
When no rate card exists, the quote is the only artefact you can compare, so make it complete before you compare it. Ask for the all-in cost per minute at your forecast volume, what happens above and below that forecast, what telephony and integration work is billed separately, and what the renewal uplift is capped at.
- All-in cost per minute at your actual forecast volume, not a headline tier
- What is excluded: telephony, integration build, professional services
- Renewal terms and any cap on annual increases
- What you keep if you leave: recordings, transcripts, configuration
Why the opacity is not automatically a problem
Enterprise contact-centre deployments genuinely vary by an order of magnitude depending on languages, integrations, concurrency and compliance, and a published rate card would mislead more than it helped. The problem is not that PolyAI quotes; it is that you cannot benchmark a quote you cannot compare. Get two other quotes on the same scope.
The alternative for teams that want a fixed cost
If the reason you are evaluating an enterprise platform is data control or predictable cost rather than scale for its own sake, an on-premise deployment answers both directly. Centricall licenses its platform for deployment inside your own cloud account or data centre, so call recordings and caller data never leave your environment and there is no per-minute platform meter at all.
Frequently asked questions
How does PolyAI pricing compare with other AI agent pricing models?
- PolyAI sits at the enterprise end of the market, and Poly AI pricing is quoted rather than published, which is normal at that tier and makes a like-for-like comparison genuinely hard. Most conversational AI pricing in the mid-market is per minute with a platform fee on top, while enterprise AI pricing is usually an annual commitment with a minimum volume attached and a rate that improves as you clear it. The only honest way to compare the two shapes is to normalise everything to all-in cost per minute at the volume you actually expect, including telephony, and to ask what the rate does in month thirteen.
How much does PolyAI cost per minute?
- PolyAI does not publish a per-minute rate. Pricing is quoted per client through a sales process and billed on usage, with no published rate card, no self-service signup and no free tier. Verified August 12, 2026.
Why does PolyAI not publish pricing?
- Enterprise voice deployments vary enormously by language coverage, integration depth, concurrency and compliance obligations, so a single published rate would be misleading for most buyers. The trade-off is that you cannot benchmark it without running a competitive process.
What should I compare an enterprise voice AI quote against?
- Normalise everything to all-in cost per minute at your forecast volume, then add the excluded items: telephony, integration build, professional services and compliance add-ons. Compare that against a published per-minute platform and against a one-time license, which removes the per-minute meter entirely.
Are the PolyAI cost estimates circulating online reliable?
- Treat them cautiously. Various third-party figures are quoted for its entry point, none confirmed by the vendor, and most originate on competitors' comparison pages. We do not repeat them here for that reason. Ask for a quote against your own scope.
Is there an option with no per-minute cost at all?
- Yes. A one-time on-premise or white-label license replaces the per-minute meter with a single purchase, after which you pay only your own hosting, telephony and model usage at provider rates. That suits organisations with high call volumes or strict data-residency requirements.
Sources
- PolyAI — https://poly.ai/ · verified August 12, 2026
- Vapi — https://vapi.ai/pricing · verified August 12, 2026
- Retell AI — https://www.retellai.com/pricing · verified August 12, 2026
Competitor pricing changes frequently. Every figure above was read from the vendor’s own published page on the date shown. Verify before quoting a client. Product names are the trademarks of their respective owners and are used here for identification only.
Further reading
Read next
More comparisons
White label platforms
The agency tiers compared on what they actually publish: monthly cost, included minutes, client account limits and rebilling rights.
AI receptionist vs answering service
Live answering services publish rates that work out at several dollars a minute. Here is the arithmetic, and what each option actually completes.
Vapi vs Retell AI
Two developer-first voice platforms, compared on published per-minute pricing, what each figure includes, and what HIPAA costs on top.
Get started
See it on your own calls
We publish what we charge and quote against your call volume, not against a discovery process. Book a call and see the numbers.
Book a discovery call