AI voice agent pricing per minute: the real all-in cost
Short answer: On one standard setup (a mid-priced OpenAI model, a standard voice, a US number), Vapi costs $0.082 to $0.129 a minute all-in, Retell AI $0.117 and Bland AI $0.14 plus telephony, as of 14 Nov 2026. ElevenLabs ($0.08) and PolyAI ($0.095) start lower but don't publish every component. Flat-plan receptionists run from $0.15 a minute (Rosie) to $2.00 a call (Smith.ai) when you use the whole plan.
Every figure below comes from official pricing pages and docs, re-checked on 14 Nov 2026. Prices change often, so check the vendor's page before you buy.
How much does every platform cost? (as of 14 Nov 2026)
| Platform | How it bills | Headline price | All-in or effective cost | Watch out |
|---|---|---|---|---|
| Retell AI | Per minute, itemized | $0.07 to $0.31/min | $0.117/min on our setup | $2/month per number |
| Vapi | Hosting fee + providers at cost | $0.05/min hosting | $0.082 to $0.129/min | Free number is inbound-only |
| Bland AI | Per minute, bundled | $0.14/min (Start) | $0.14/min + telephony | Telephony rate not published |
| ElevenLabs Agents | Plan minutes | $6 to $990/month | $0.08/min + LLM + telephony | LLM and telephony not priced per minute |
| PolyAI Platform | Per minute, bundled | 9.5¢+/min | $0.095/min + telephony | Telephony not priced |
| PolyAI Model API | Per minute + text tokens | 2¢/min | Not computable | Token prices not published |
| Goodcall | Per unique customer | $79/month for 100 | $0.15 to $0.79 per customer | Extra customers 15¢ to 79¢ |
| Smith.ai | Per call | $150/month for 75 | $1.67 to $2.00 per call | Extra calls $2.17 to $3.00 |
| Rosie | Minutes | $49/month for 250 | $0.15 to $0.20/min | Overage rate not published |
| Upfirst | Per call | $24.95/month for 30 | $0.50 to $0.83 per call | Extra calls 70¢ to $1.50 |
| Dialzara | Minutes | $29/month for 60 | $0.35 to $0.48/min | Overage 35¢ to 48¢/min |
| Phonely | Minutes | $50/month | $0.20 to $0.25/min | Page gives 2 minute counts per plan |
| My AI Front Desk | Minutes + credits | $99/month for 200 | $0.495/min | Included minutes cost more than overage |
| Slang.ai | Per location | From $399/location | Not computable | Billing period and minutes not stated |
| Synthflow | Annual enterprise contract | Not published | Not published | Sales only, no trial |
| Thoughtly | Sales only | Not published | Not published | No public price |
How did we price a "standard setup"?
To compare like with like, Kall priced one setup on each per-minute platform: - Inbound calls to a US number on the platform's own phone service. - A mid-priced OpenAI model. On Retell that's GPT 6.1 Terra, one of 3 models Retell tags "Recommended", at $0.032/min on the standard tier ($0.064 on its fast tier). - The platform's standard voice, not a premium one. - Pay-as-you-go, no add-ons, no volume discount.
Where a vendor publishes only a range, we show the range. Where it publishes nothing, we write "not published" and leave it out of the total, so those totals are floors, not estimates.
What does each per-minute platform charge for each part of a call?
| Component | Retell AI | Vapi | Bland AI | ElevenLabs Agents | PolyAI Platform | PolyAI Model API |
|---|---|---|---|---|---|---|
| Platform | $0.055 | $0.05 | $0.14 (bundle) | $0.08 (bundle) | $0.05 | $0.02 (model) |
| LLM | $0.032 | $0.0077 to $0.0452 | In bundle | Not published per minute | $0.03 (Dialog-RSN-1) | Text tokens: not published |
| Voice (TTS) | $0.015 | $0.0146 to $0.0238 | In bundle | In bundle | $0.015 | Not published |
| Speech-to-text | No separate charge listed | $0.0095 to $0.0099 | In bundle | In bundle | No separate charge listed | No separate charge listed |
| US telephony | $0.015 | $0 (free Vapi number) | Not published | Not published ("at cost") | Not published | Not published |
| All-in per minute | $0.117 | $0.0818 to $0.1289 | $0.14 + telephony | $0.08 + LLM + telephony | $0.095 + telephony | $0.02 + unpublished parts |
The arithmetic, as of 14 Nov 2026: - Retell AI: $0.055 Voice Infra + $0.032 LLM + $0.015 voice + $0.015 telephony = $0.117/min. 1,000 minutes cost $117, plus $2 for the number. Retell bills to the second. - Vapi: $0.05 hosting + $0.0077 OpenAI + $0.0146 ElevenLabs voice + $0.0095 Deepgram transcription + $0 = $0.0818 at the low end; $0.05 + $0.0452 + $0.0238 + $0.0099 = $0.1289 at the high end. Vapi's own estimate for 1,000 minutes is $82 to $129. It lists OpenAI as a range, not per model, so we can't isolate one mid-priced model. - Bland AI: "One flat rate covers the LLM, speech-to-text, and text-to-speech" at $0.14/min. Telephony comes "on your own carrier or Bland's at pass-through cost", with no rate given. The $299/month Build plan cuts talk time to $0.12, which only pays off above 14,950 minutes a month ($299 ÷ $0.02). - ElevenLabs Agents: every paid plan works out to $0.08/min (Pro: $99 ÷ 1,238 = $0.080), and extra minutes cost $0.08. LLM tokens pass through "at the provider's published rate, with no markup" (ElevenLabs docs), though the same page says some Gemini and Claude models bill at Vertex AI regional rates, 10% higher on tokens. Telephony is "at cost". Neither the LLM nor telephony has a per-minute figure. - PolyAI Platform: 5¢ platform + 3¢ Dialog-RSN-1 + 1.5¢ standard voice = 9.5¢ ($0.095)/min. Telephony isn't priced. - PolyAI Model API: 2¢/min "plus text tokens", for "developers running their own voice stack". The page gives no token rate and no voice or telephony price, so there's no all-in figure to compute.
Choosing between the two closest rivals? See Retell AI vs Vapi: pricing, compliance and fit compared. If tracking Vapi's per-provider charges is the problem, see Best Vapi alternatives.
What do flat-plan AI receptionists really cost per minute or call?
Receptionist apps sell a monthly bundle. We divided each plan's price by what it includes (monthly billing, as of 14 Nov 2026). That's the best case: use half the bundle and your effective rate doubles.
| Plan | Price/month | Included | Effective cost at the cap | Overage |
|---|---|---|---|---|
| Goodcall Starter | $79 | 100 unique customers | $79 ÷ 100 = $0.79/customer | $0.79 |
| Goodcall Scale | $299 | 2,000 unique customers | $0.15/customer | $0.15 |
| Smith.ai Pro | $150 | 75 calls | $150 ÷ 75 = $2.00/call | $2.50 |
| Smith.ai Enterprise | $500 | 300 calls | $1.67/call | $2.17 |
| Rosie Professional | $49 | 250 min | $49 ÷ 250 = $0.20/min | Not published |
| Rosie Scale | $149 | 1,000 min | $0.15/min | Not published |
| Upfirst Starter | $24.95 | 30 calls | $24.95 ÷ 30 = $0.83/call | $1.50 |
| Upfirst Scale | $299 | 600 calls | $0.50/call | $0.70 |
| Dialzara Business Lite | $29 | 60 min | $29 ÷ 60 = $0.48/min | $0.48 |
| Dialzara Business Elite | $349 | 1,000 min | $0.35/min | $0.35 |
| Phonely Starter | $50 | 200 or 250 min | $0.25 or $0.20/min | $0.25 |
| Phonely Pro | $150 | 650 or 750 min | $0.23 or $0.20/min | $0.30 |
| My AI Front Desk | $99 | 200 min | $99 ÷ 200 = $0.495/min | $0.25 |
| PolyAI Essentials | $49 | 250 min | $0.20/min | $0.30 |
| Slang.ai Core | $399 per location | Not stated | Not computable | Not stated |
What the table doesn't show: - Per-call plans ignore call length. Smith.ai and Upfirst charge per call, and Goodcall counts each caller once a month however often they ring. - Spam is mostly free. Upfirst doesn't bill calls under 15 seconds or spam, Goodcall doesn't count spam or wrong numbers, and Smith.ai filters known spammers before answering. - My AI Front Desk runs backwards. An overage minute (25 credits at $0.01) costs $0.25, about half what an included minute works out to. The plan also includes 1,000 overage credits a month. - Phonely's page disagrees with itself. Plan cards say 200 and 650 minutes; the comparison table says 250 and 750. Ask before you buy. - Annual billing lowers the numbers. Upfirst Scale drops to $240/month ($0.40/call) and Phonely Pro to $100/month ($0.13 to $0.15/min).
For which receptionist fits which business, see Best AI receptionists for small businesses.
What hidden costs push the bill up?
These extras sit outside the per-minute and plan prices above (all as of 14 Nov 2026).
Phone numbers - Retell: $2/month per number. Its pricing page lists a verified number at "$10.00/Phone number/month" in one place and as a "One-time $10 fee" in another, so ask Retell which applies. Branded caller ID adds $0.10 per outbound call. - Vapi: 1 free US number, inbound only. Outbound calls need a number imported from another provider (Vapi docs). - Bland: the free Start plan includes an inbound number it values at $15/month. - Included on Goodcall, Upfirst and PolyAI's plans. Phonely includes 3 (Starter) or 5 (Pro).
Overages: $0.20 to $0.48 per extra minute on minute-based plans and $0.70 to $3.00 per extra call on per-call plans. Rosie doesn't publish its rate.
HIPAA add-ons - Vapi: $2,000/month, including a DPA and BAA (Vapi), switched on in the dashboard (Vapi docs). - Retell: its docs say you self-sign the BAA and "there is no additional fee to sign these agreements", without naming a plan (Retell docs). Its pricing page also has "HIPAA / BAA" and "Custom BAA" rows in a table comparing Pay-as-you-go with Enterprise, so confirm with Retell which plan you need. - Enterprise only, at a custom price: ElevenLabs, PolyAI, Upfirst and Thoughtly. Bland's FAQ says it signs a BAA, but its plan table lists the BAA under Enterprise only. Phonely's FAQ says Enterprise; its comparison table is unclear. - Not mentioned on the pricing pages of Goodcall, Smith.ai, Rosie, Dialzara, My AI Front Desk or Slang.ai. Healthcare buyers should start with Best AI receptionists for dental and medical clinics.
Concurrency (calls at the same time) - Retell: 20 free, then $8 per line per month. - Vapi: 4 on pay-as-you-go, 10 on Core ($29/month), 30 on Pro, then $10 per line per month. - Bland: 10 on Start (plus a cap of 100 calls a day), 50 on Build. - ElevenLabs: 4 on Free up to 40 on Business. PolyAI: 5 to 10 by plan. Phonely: unlimited.
Minimums and billing rules - Vapi Pro charges 10% of the hosting fee with a $999/month minimum. Bland Build is $299/month. - Retell bills calls under 10 seconds with a dynamic opening message as 10 seconds, bills silence and hold time, and bills more time when the average prompt tops 4,000 tokens (4,800 tokens = 20% more). - ElevenLabs bills silence at 5% of the per-minute rate (ElevenLabs docs). - Bland bills transfer time at $0.05/min on Start and $0.04 on Build.
Which AI voice platforms don't publish prices?
- Synthflow sells "an annual enterprise agreement" only, with "no free plan or trial and no rate by the minute" (Synthflow). See Synthflow alternatives after its move to enterprise-only.
- Thoughtly shows no prices. Every plan says "Talk to sales", "Get a quote" or "Talk to enterprise" (Thoughtly).
- Slang.ai shows starting prices per location, but every plan button says "Get a quote".
- Enterprise tiers at Retell, Vapi (Premier), Bland, ElevenLabs and PolyAI are custom.
How we calculated these numbers
- Official pricing pages and docs only, re-checked on 14 Nov 2026.
- One standard setup for per-minute platforms, with no estimates for unpublished parts.
- Flat plans divided at the plan cap, on monthly billing.
- Research only: we haven't made test calls yet. Method: How Kall tests and ranks AI voice agents.
Sources
- Pricing, Retell AI, 14 Nov 2026
- Compliance, Retell AI docs, 14 Nov 2026
- Pricing, Vapi, 14 Nov 2026
- Creating free phone numbers, Vapi docs, 14 Nov 2026
- HIPAA compliance, Vapi docs, 14 Nov 2026
- Pricing, Bland AI, 14 Nov 2026
- ElevenAgents pricing, ElevenLabs, 14 Nov 2026
- Supported LLMs, ElevenLabs docs, 14 Nov 2026
- Optimizing LLM costs, ElevenLabs docs, 14 Nov 2026
- Pricing, PolyAI, 14 Nov 2026
- Pricing, Goodcall, 14 Nov 2026
- AI Receptionist pricing, Smith.ai, 14 Nov 2026
- Pricing, Rosie, 14 Nov 2026
- Pricing, Upfirst, 14 Nov 2026
- Pricing, Dialzara, 14 Nov 2026
- Pricing, Phonely, 14 Nov 2026
- Pricing, My AI Front Desk, 14 Nov 2026
- Pricing, Slang.ai, 14 Nov 2026
- Pricing, Synthflow, 14 Nov 2026
- Pricing, Thoughtly, 14 Nov 2026
FAQ
How much does an AI voice agent cost per minute?
On one standard setup, developer platforms cost about $0.08 to $0.14 a minute all-in as of 14 Nov 2026: Vapi $0.082 to $0.129, Retell $0.117 and Bland $0.14 plus telephony. Premium voices and bigger models cost more.
Why is the headline per-minute price not the real cost?
Most developer platforms bill the platform fee separately from the LLM, voice, transcription and phone line. Vapi's $0.05 hosting fee becomes $0.082 to $0.129 a minute once those are added, using Vapi's own published ranges.
What is the cheapest AI receptionist per minute?
On monthly billing, Rosie's Scale and Growth plans work out cheapest among minute-based plans, at about $0.15 a minute if you use every included minute, as of 14 Nov 2026. Use half your minutes and the effective rate doubles.
What hidden costs should I budget for?
Phone numbers ($2 a month each on Retell), extra concurrent lines ($8 to $10 a month on Retell and Vapi), HIPAA add-ons ($2,000 a month on Vapi), overage rates and monthly minimums such as Vapi Pro's $999.
Does Synthflow publish its pricing?
No. As of 14 Nov 2026 Synthflow sells only annual enterprise agreements, and its pricing page shows no free plan, trial or per-minute rate. Thoughtly is also sales-only.
Related
- Retell AI vs Vapi: pricing, compliance and fit comparedRetell AI vs Vapi on per-minute cost, free credits, HIPAA and SOC 2, integrations, no-code fit and languages, from official pages as of 14 Nov 2026.
- Best Vapi alternatives7 Vapi alternatives with pricing as of 14 Nov 2026: Retell AI, Bland, ElevenLabs, Deepgram, PolyAI, LiveKit and Pipecat, with a watch-out for each.
- Best AI receptionists for small businesses8 AI receptionists for small businesses compared on price, free trial, booking, integrations, call transfer and languages, as of 14 Nov 2026.
- Best AI receptionists for dental and medical clinicsWhich dental and medical AI receptionists sign a HIPAA BAA, on which plan and at what cost, and which connect to Dentrix, Open Dental or Epic.
Ranks are never for sale. This page has no affiliate links and no paid placements. Prices change often: check the vendor's pricing page before you buy.