Contents
Fast answer
PolyAI publishes no price. When we checked on 20 August 2026, the only billing-related statement on its pricing page was that ongoing use of the voice agent is priced on a per-minute basis, and that the charge includes proactive performance improvements, maintenance and 24/7 support. No rate appears. No minimum commitment, contract length, implementation fee, trial duration or self-serve tier is published anywhere on the page, and every route leads to a demo request. That is a legitimate way to sell enterprise contact-centre software, and it also means you cannot evaluate PolyAI on cost without entering a sales process that takes weeks. If you are a small business weighing missed calls and bookings, the honest answer is that this product is not sized for you and the time spent finding its price is better spent on a platform that publishes one.
Facts last verified 2026-08
Quick facts
- Published rate
- None (checked 2026-08-20)
- What is published
- Usage is priced on a per-minute basis
- Included in that charge
- Proactive performance improvements, maintenance and 24/7 support
- Minimum commitment
- Not published
- Implementation fee
- Not published
- Free trial
- Not published
- Self-serve signup
- None; every route is a demo request
- Benchmark band
- Published voice rates elsewhere run $0.05 to $0.31 a minute
On this page
What this page covers
A comparison visitor should understand the tradeoff, the best-fit scenario, and the next diagnostic tool to confirm the choice.
- 01What is published
- 02What is not
- 03How to benchmark a quote
- 04Where the meter bites
- 05Cheapest honest path
- 06When to leave
Why does this matter now?
An unpublished price is itself information. It tells you the product is sold to buyers who expect a procurement cycle, that the number is set per customer rather than per plan, and that the vendor expects the implementation to be scoped. None of that is dishonest, but all of it costs you calendar time before you learn anything. Knowing what to ask for in the first call, and what to measure the answer against, turns a multi-week evaluation into a single conversation.
Internal path
Where to go next from this page
These links are part of the A8gent learning and conversion path. Use them to move from concept, to diagnosis, to workflow build, to course.
What you should be able to do after this
- Know exactly what PolyAI does publish
- Ask for the four numbers that decide viability
- Benchmark a quote against published rates
- Decide whether to enter the sales process at all
How do you do it, step by step?
1. Note precisely what is published
One sentence: usage is priced per minute, and that charge includes proactive performance improvements, maintenance and 24/7 support. That is genuinely useful, because it tells you the model is per-minute usage rather than per-seat or per-resolution, and that ongoing optimisation is bundled rather than billed as professional services.
2. Ask for four numbers, separately, in the first call
The per-minute rate. The minimum commitment, in both spend and term. The implementation or build fee as its own line. And the renewal mechanics. Ask for all four in writing. The one that most often surfaces last is the minimum commitment, and it is usually the one that decides whether the deal is viable at all.
3. Benchmark the quoted rate against published ones
You cannot judge a per-minute rate in isolation. Published August 2026 comparisons: Bland bundles at $0.14 a minute with no monthly fee, Retell publishes an itemised $0.07 to $0.31, ElevenLabs Agents works out at $0.08 across its tiers, and Vapi charges $0.05 for orchestration with providers separate. A quote well above that band needs to be earning it.
4. Price the implementation as its own decision
Enterprise voice deployments of this kind are scoped and built rather than configured, and that work is billed separately from usage. PolyAI publishes no figure for it. Insist on seeing it as a standalone number rather than folded into a blended annual figure, because it is the part that does not recur and therefore the part that should be negotiated hardest.
5. Decide whether you are the buyer this is built for
There is no self-serve tier and no trial, so you cannot hear it working on your own calls quickly. For a contact centre with hundreds of agents and a procurement function, that is normal. For a clinic, a trades business or an agency, it is weeks of calendar time to reach a number that a competitor puts on a public page.
What an unpublished price actually tells you
PolyAI publishes one sentence about money. Ongoing use of the voice agent is priced on a per-minute basis, and that charge includes proactive performance improvements, maintenance and 24/7 support. That is the whole of the public information as of 20 August 2026. No rate, no minimum commitment, no contract length, no implementation fee, no trial duration and no self-serve tier appear anywhere on the page, and every route leads to a demo request.
That single sentence is more useful than it first appears, and it is worth extracting everything from it before concluding the page says nothing. It tells you the commercial model is usage-based rather than per-seat, which means your bill scales with call volume rather than with headcount. It tells you that ongoing optimisation and support are inside the rate rather than billed as professional services, which is a genuine difference from vendors who charge separately for tuning after go-live.
The absence is informative too. An unpublished price signals that the number is set per customer rather than per plan, that the vendor expects a procurement conversation, and that implementation is scoped rather than configured. None of that is dishonest and all of it is normal for enterprise contact-centre software. What it does mean is that you cannot budget, seek board approval or run a vendor comparison without first entering a sales process, and that process is measured in weeks.
So the decision to make before anything else is whether you are the buyer this is built for. A contact centre with hundreds of agents, a procurement function and a compliance review has the machinery to absorb that process and will find the outcome reasonable. A clinic, a trades business or an agency comparing per-minute rates does not, and the weeks spent finding out are weeks during which the phone is still going unanswered. That is the real cost of an unpublished price and it is not denominated in dollars.
The four questions to ask, and the order to ask them
If you do enter the process, go in with four specific numbers to extract, in writing, and separately rather than as a blended annual figure. The per-minute rate. The minimum commitment, in both spend and term. The implementation or build fee as its own line. And the renewal mechanics, specifically what happens to the rate when the initial term ends. Asking for all four upfront is normal, and a vendor unwilling to separate them is telling you something useful.
Order matters. The minimum commitment is the one that most often decides viability and most often surfaces last, typically after you have invested several calls and a technical review. Asking for it early is not aggressive, it is efficient for both sides, because a floor above your budget ends the conversation in week one rather than week six. Ask what the minimum is in annual spend and what the shortest available term is, and treat a reluctance to answer as an answer.
The implementation fee deserves to be isolated because it does not recur and therefore behaves differently in negotiation. Enterprise voice deployments of this kind are scoped and built rather than configured, and that work is real and worth paying for. Folded into a blended per-minute figure across a multi-year term it becomes invisible and unnegotiable. Broken out, it is a one-off number you can compare against what the same build would cost elsewhere, including building it yourself.
Two further questions are worth adding because the pricing page does not answer them and they affect reliability rather than cost. What concurrency is included, and what happens when it is exceeded? And is there a published status page or uptime commitment? Neither concurrency figures nor rate limits nor a status page were published where we could find them, which means reliability cannot be assessed independently and has to be established contractually.
Benchmarking a quote you finally receive
A per-minute rate in isolation is unjudgeable, so bring published comparators to the conversation. Read on 20 August 2026: Vapi charges $0.05 a minute for orchestration with the model, speech and telephony billed by your own providers or at nothing to Vapi with your own keys. ElevenLabs Agents works out at exactly $0.08 a minute across every tier, with the model and telephony separate. Retell AI publishes an itemised $0.07 to $0.31 a minute including the model and US telephony. Bland AI bundles at $0.14 a minute with no monthly fee.
Those four span the honest range for self-serve voice agents, roughly five to thirty-one cents a minute depending on what is included and which model you pick. A PolyAI quote materially above that band is not automatically wrong, because it includes things those platforms do not: scoping, build, ongoing optimisation and 24/7 support inside the rate. But it does need to be earning the difference, and framing the conversation that way is far more productive than asking whether the price is good.
Do the arithmetic on your own volume before the call so the comparison is concrete. At 1,500 minutes a month, the reference workload for a busy clinic or trades business, those published platforms land between roughly $75 and $465 a month all-in depending on configuration. If a quote arrives at several times that, the question is what the additional spend buys and whether you would otherwise have paid for it separately in staff time, agency fees or a failed self-serve implementation.
Compare the support model honestly as well, because it is the strongest part of the offer. On Vapi or LiveKit, a latency problem at nine on a Monday morning is yours to diagnose across four providers. PolyAI's rate explicitly includes maintenance and 24/7 support. For an organisation without engineering capacity, that difference is not a line item, it is the entire reason the category exists, and it deserves to be weighed against the rate rather than separately from it.
At a glance
PolyAI published pricing information, read from poly.ai/pricing on 20 August 2026
| Question a buyer needs answered | What PolyAI publishes | What to do |
|---|---|---|
| What is the per-minute rate? | Not published | Ask for it in writing in the first call |
| How is usage billed? | Per minute, including improvements, maintenance and 24/7 support | Confirm what counts as a billable minute |
| What is the minimum commitment? | Not published | Ask for spend and term separately; this decides viability |
| What does implementation cost? | Not published | Insist on it as a standalone line, not folded into usage |
| Is there a trial? | Not published | Expect to evaluate through a sales process rather than your own calls |
| Is there self-serve signup? | No | Budget calendar time before you learn any number |
| What is concurrency? | Not published | Ask, and ask what happens when it is exceeded |
| Benchmark: Bland AI | $0.14/min bundled, no monthly fee | Published on the vendor's own page |
| Benchmark: Retell AI | $0.07 to $0.31/min itemised | Published on the vendor's own page |
| Benchmark: ElevenLabs Agents | $0.08/min effective across tiers | Published on the vendor's own page |
| Benchmark: Vapi | $0.05/min orchestration, providers separate | Published on the vendor's own page |
What mistakes should you avoid?
- Treating an unpublished price as a signal of quality rather than of sales motion
- Accepting a blended annual figure that hides the implementation fee inside usage
- Leaving the minimum commitment question until late in the process
- Evaluating without a benchmark from platforms that do publish per-minute rates
- Entering a months-long procurement cycle for a workload a published-price platform handles
- Assuming reliability can be assessed independently when no concurrency figures or status page are published
FAQ
- How much does PolyAI cost?
- Not published. When we checked on 20 August 2026 the pricing page stated only that ongoing use is priced per minute, with proactive performance improvements, maintenance and 24/7 support included in that charge. No rate, minimum, implementation fee or trial length appears anywhere on the page.
- Is there a free trial or self-serve tier?
- No. There is no published trial and no self-serve signup. Every route on the pricing page leads to a demo request, so evaluating PolyAI means entering a sales process rather than testing it on your own calls in an afternoon.
- What should I ask for in the first sales call?
- The per-minute rate, the minimum commitment in both spend and term, the implementation or build fee as its own line, and what happens at renewal. Get all four in writing. The minimum commitment tends to surface last and is usually the number that decides viability.
- What should I compare a PolyAI quote against?
- Published August 2026 per-minute rates from platforms that print them: Bland at $0.14 bundled with no monthly fee, Retell's itemised $0.07 to $0.31 band, ElevenLabs Agents at an effective $0.08 across its tiers, and Vapi at $0.05 for orchestration with model, speech and telephony billed separately.
- Is PolyAI right for a small business?
- Rarely. It is enterprise contact-centre software sold through a sales process, with no published price, no trial and procurement measured in weeks or months. A small business handling missed calls and bookings will get to a working agent faster and cheaper on a platform with a published rate and a self-serve signup.
Sources & further reading
Done for you
Want this built for your business?
Tell us the workflow. We scope, build, and ship the agent with guardrails and the numbers to prove it worked. The scoping call is free.
Free scoping call. You own the code.
Was this page helpful?

