Published entry prices for AI answering services run from nothing to £399 a month, and two suppliers with near-identical marketing can differ by a factor of ten on the same volume of calls.

The spread has almost nothing to do with quality. It comes down to what a supplier counts, and what happens when you go over. Every figure below was taken from suppliers' own pricing pages in August 2026.

The four billing models

Before comparing prices, work out which unit you are buying.

Per call. You pay for each answered call, whatever its length. A two-minute call and a six-minute call cost the same. Smith.ai and Penelope both work this way.

Per minute. You buy a bundle of minutes and pay an overage rate beyond it. Dialzara, Phonely and Bland work this way. Long calls cost more.

Per unique caller. Goodcall counts distinct phone numbers each month. If the same client rings four times, that is one billable caller.

Per location, flat fee. Slang.ai charges a set monthly fee per site with no call bundle. Predictable, and expensive if the site is quiet.

Suppliers disagree about this in public, which is useful to know. Smith.ai's own FAQ argues that per-minute billing sets a supplier's incentives against yours, because a per-minute supplier profits from longer calls. Goodcall makes a similar argument for counting callers. Both have a point, and both chose the model that flatters their own numbers. Ours bills per call, for the reason Smith.ai gives.

What suppliers charge

Entry points, checked August 2026:

  • Smith.ai: free for 25 answered calls, then $3.00 a call. First paid tier $150 a month for 75 calls. Human-answered service is separate, from $300 a month for 30 calls at $11.50 a call beyond that.
  • Dialzara: $29 a month for 60 minutes, then $0.48 a minute.
  • Rosie: $49 a month for 250 minutes. Overage rate not published.
  • Penelope: £49 a month for 25 answered calls, extra calls £2.16. £179 for 100 calls, £399 for 250. Guided setup included, no setup fee.
  • Phonely: free for 100 minutes, then $50 a month, at $0.25 to $0.35 a minute.
  • Goodcall: $79 a month per agent, unlimited minutes, 100 unique callers, then $0.50 per extra caller.
  • Frontdesk: voice starts at $99 a month for 200 minutes.
  • Slang.ai: $399 a month per location, restaurants only.
  • Bland: $0.14 a minute, no platform fee on the entry tier.
  • Synthflow: enterprise contracts from $30,000 a year.
  • RingCentral: no published price.

Three of those deserve a warning. Frontdesk's widely quoted $20 plan includes no voice minutes at all, only chat, SMS and email drafts. Synthflow has withdrawn self-serve pricing entirely, so any article quoting it at $29 to $450 a month is out of date. And Phonely's own page contradicts itself, listing different minute allowances and two different overage rates depending on which table you read.

What the headline price leaves out

Overage. This is where budgets break. Dialzara's entry plan gives 60 minutes for $29, but at $0.48 a minute a firm taking 300 minutes of calls pays $144, not $29. Price your actual volume, never the entry tier.

Spam. If a supplier bills per call, recorded sales calls count unless they are filtered out first. Ask what happens to them.

Transfers and post-call time. Bland charges separately for transfer minutes. Phonely bills post-transfer time. Small individually, and they accumulate.

Telephony. Developer platforms quote the AI cost only. Bland's per-minute rate covers the language model, speech recognition and voice synthesis; phone carriage is billed separately.

Add-ons priced per event. Smith.ai's human service charges $1.50 per appointment booked, $0.50 per notification and $0.25 per recording. A firm booking 30 appointments a month adds $45 to a $300 plan.

Multipliers. Goodcall charges per agent, Slang.ai per site. Three locations on Slang.ai's entry tier is $1,197 a month.

Setup. Some suppliers charge for it. Some hand you a dashboard and leave you to configure it yourself, which is a cost measured in your time rather than theirs.

Working out your real monthly cost

Take your own numbers rather than a supplier's example.

Count the calls you get in a month, splitting office hours from out of hours. Estimate average call length: two to three minutes is typical for a booking or enquiry, longer for anything involving advice. Multiply. Then price that volume against each model, and add the extras that apply to you.

Two patterns emerge from doing this properly. Short, high-volume calls favour per-minute pricing. Longer calls, or calls where you want the caller kept on the line until something is resolved, favour per-call pricing.

Is it cheaper than employing someone?

A receptionist on £25,000 costs an employer roughly £29,000 to £30,000 once National Insurance and pension contributions are added, before holiday cover, sick pay, a desk and a phone. That is around £2,400 a month for one person covering about 40 hours of a 168-hour week.

An AI service covering the whole week runs from £49 a month. On cost per hour covered, it is not a close contest.

What that comparison hides is that the two are not doing the same job. An in-house receptionist greets visitors, chases suppliers, and knows that the caller with the difficult name is the client who nearly left last year. If those things matter, the honest comparison is between an AI service and voicemail, not between an AI service and your receptionist.

What to check before you sign

Price your actual monthly volume including overage. Confirm the billing unit. Ask what counts as a billable call and how spam is treated. Get transfer and booking charges in writing. Check the minimum term and the notice period: ours is a three-month minimum on Starter, Growth and Pro, then one month's notice. And ask what happens to your number and your call data when you leave.

Suppliers change prices often, and several figures here contradict what third-party comparison articles still publish. Take pricing from the supplier's own page on the day you decide, and get your quote in writing.

We set this up with you

Tell us about your firm and we’ll build Penelope around it.

Book a demo