Quick answer: ElevenLabs Agents bills $0.080 per call minute including speech to text and voice; Vapi bills $0.05 hosting plus each provider at cost. For 10,000 US outbound minutes on Twilio we get $988.85–$1,460.15 on Vapi and $964.92–$1,387.19 on ElevenLabs with GPT-5 Mini to GPT-5.5 ($945.87–$1,803.15 across the OpenAI models it lists), so concurrency, batch limits and HIPAA terms decide it (US dollars, as at 4 October 2026).
“I need my agent to make and take real phone calls — Vapi or ElevenLabs?” Both are developer platforms, and both put a Twilio number or a SIP trunk between your agent and the phone network. They differ in how the minute is metered, how many calls can run at once, what their bulk-dialling tools allow and what you sign for regulated data. Every figure below was read on Vapi’s, ElevenLabs’ and Twilio’s own pages on 4 October 2026.
Vapi vs ElevenLabs Agents: what each one is on a phone line
Vapi is an orchestration layer. Its pricing page answers “What exactly is in the per-minute cost?” with: “Four components: speech-to-text, the language model, text-to-speech, and transport. Each is billed at the provider’s listed price with no markup. Vapi’s $0.05/min hosting fee is added on top.” You pick the transcriber, model and voice, and the carrier bills the call minutes.
ElevenLabs Agents, branded ElevenAgents on its own site, bundles more into one meter. Its agents pricing page lists $0.080 per additional call minute and says “Additional call minutes cost $0.08 per minute … LLM usage is billed separately on top, based on the model you choose.” Speech to text and text to speech sit inside that minute. On telephony the same page says “ElevenLabs adds no telephony fee. You connect your own provider, such as Twilio or a SIP trunk, and that provider bills you directly for the carrier side of the call at whatever rate you have with them.”
One thing that surprises people: the voice can be identical on both. Vapi’s calculator lists ElevenLabs as a voice provider at $0.0146–$0.02383 a minute. So the choice is rarely about how the agent sounds; it is about the meter, the capacity and the paperwork.
Five criteria that decide Vapi vs ElevenLabs for phone calls
- The meter. One bundled rate (ElevenLabs) against a hosting fee plus pass-through lines (Vapi).
- Peak simultaneous calls. Both cap concurrency by plan and price the step-ups very differently.
- The bulk-dialling tool. Vapi Campaigns and ElevenLabs Batch Calling draw on concurrency differently.
- Data retention and HIPAA. Retention defaults, zero-retention modes and BAA availability differ by plan.
- Who controls the stack. Vapi lets you swap each component; ElevenLabs bundles speech and lets you choose the model.
Criteria 2, 3 and 4 are where the platforms diverge for a calling programme; the worked bills show criterion 1 matters less than it looks.
How Vapi and ElevenLabs Agents charge for a call minute
Component by component, as published on 4 October 2026, in US dollars. A Vapi cell that shows a band is Vapi’s own calculator band for that provider, not a price per model.
| Component | Vapi | ElevenLabs Agents |
|---|---|---|
| Platform fee per minute | $0.05 hosting | $0.080 per call minute beyond the plan’s included minutes |
| Speech to text | Pass-through; Deepgram $0.00953–$0.00994, AssemblyAI $0.00521–$0.00747, OpenAI $0.0118–$0.01221 | Included in the call minute |
| Voice (text to speech) | Pass-through; ElevenLabs $0.0146–$0.02383, Vapi voices $0.00695–$0.01934, Cartesia $0.00689–$0.01294 | Included in the call minute |
| Language model | Pass-through, no markup; OpenAI $0.00774–$0.04523, Anthropic $0.00818–$0.03656, Google $0.006–$0.03329 | “LLM at provider cost”; calculator estimate per model, e.g. GPT-5 Mini $0.00238, GPT-4.1 $0.01664, GPT-5.5 $0.04461, Claude Sonnet 4.5 $0.02845, Gemini 2.5 Flash $0.00289; select Gemini and Claude models at Vertex AI regional rates, “a 10% increase” |
| Carrier (US) | “Charged by provider, not Vapi”: Twilio outbound $0.014, Twilio inbound $0.008, Vonage $0.00814, Telnyx $0.0055; Vapi Telephony / SIP free | No ElevenLabs telephony fee; Twilio or your SIP provider bills you directly |
| Monthly fee | Usage only $0; Core $29; Pro 10% of hosting fees, $999 minimum; Premier custom | Free $0, Starter $6, Creator $22, Pro $99, Scale $299, Business $990; Enterprise custom |
| Concurrent calls included | 4 (Usage only), 10 (Core), 30 (Pro), custom (Premier) | 4, 6, 10, 20, 30, 40 (Free to Business); Enterprise “elevated” |
| More concurrency | $10 per line per month, on any package | Burst up to 3× the limit (capped at 300 for non-enterprise) at $0.160 per minute; or Enterprise |
| Silence on the line | “Every minute your assistants run is billed”; voice billed per character spoken | Silence longer than 10 seconds billed at a 95% discount |
| Free start | $5 free credits; 1 free US number (inbound only) | Free plan with 15 minutes of calls |
Add Vapi’s three speech and platform lines on the ElevenLabs-voice stack and you get $0.05 + $0.00953 + $0.0146 = $0.07413 a minute at the low end and $0.05 + $0.00994 + $0.02383 = $0.08377 at the high end. ElevenLabs’ $0.080 sits inside that band. Before the model and the carrier, the two meters are within $0.006 a minute of each other.
The $0.08 rule: ElevenAgents plans buy concurrency, not cheaper minutes
Divide each ElevenAgents plan price by the $0.08 additional-minute rate and you get that plan’s included minutes. The plans are prepaid minutes at the overage rate; what changes between them is how many calls can run at once.
| Plan | Monthly price | Price ÷ $0.08 | Included minutes (published) | Concurrent calls |
|---|---|---|---|---|
| Starter | $6 | 75 | 75 | 6 |
| Creator | $22 | 275 | 275 | 10 |
| Pro | $99 | 1,237.5 | 1,238 | 20 |
| Scale | $299 | 3,737.5 | 3,738 | 30 |
| Business | $990 | 12,375 | 12,375 | 40 |
The practical rule: if your monthly minutes are at or above a plan’s included minutes, moving up to that plan costs you nothing extra and buys you its concurrency. At 10,000 minutes a month, Creator, Pro and Scale all cost $800.00 or $799.96, but Scale gives you 30 simultaneous calls instead of 10. Business is the exception at that volume: $990 buys 12,375 minutes, so 2,375 of them go unused.
Vapi has no equivalent. Core is a flat $29, and each extra line is $10 a month. That grows linearly; our guide to scaling AI calling from 100 to 10,000 calls covers how to size peak concurrency from call volume and calling window.
What do 10,000 outbound minutes a month cost on Vapi vs ElevenLabs?
This is the worked calculation. Every line is a published rate multiplied by a quantity you control, so substitute your own numbers.
Scenario A assumptions: 10,000 connected minutes a month, all US to US, outbound. Peak of 10 simultaneous calls. One Twilio local number at Twilio’s own $1.15 a month on both platforms, because Vapi’s free number cannot dial out and ElevenLabs uses your carrier’s number. Twilio outbound at $0.0140 a minute on both, from Twilio’s US pricing page. Vapi stack: Deepgram transcription, an OpenAI model, an ElevenLabs voice, Core at $29 for 10 lines. ElevenLabs stack: the Scale plan at $299 (3,738 minutes, 30 calls), extra minutes at $0.080, and two named OpenAI models from ElevenLabs’ calculator: GPT-5 Mini ($0.00238 a minute) and GPT-5.5 ($0.04461 a minute). These are two points, not ElevenLabs’ range: the same calculator prices the OpenAI models on its Models page from $0.00048 a minute (GPT-5 Nano) to $0.0862 (GPT-6 Astra), which would put Scenario A on ElevenLabs anywhere from $945.87 to $1,803.15.
The LLM rows are estimates, and the two vendors estimate differently. Vapi’s calculator gives one band per provider without naming a model: $0.00774–$0.04523 a minute for OpenAI. ElevenLabs’ calculator gives one figure per named model, and its footnote calls them “Estimated costs in USD, covering ElevenAgents platform voice and LLM models.” The two do not line up: ElevenLabs’ GPT-5 Mini figure, $0.00238, sits below the bottom of Vapi’s entire OpenAI band. Vapi’s band may start from a dearer model or assume more tokens a minute; neither calculator states its tokens-per-minute assumption beside the figures. So the LLM row compares two vendors’ estimates, not two invoices.
| Line item (Scenario A) | Vapi (low) | Vapi (high) | ElevenLabs (low) | ElevenLabs (high) |
|---|---|---|---|---|
| Platform: 10,000 × $0.05 (Vapi); Scale $299 + 6,262 × $0.080 (ElevenLabs) | $500.00 | $500.00 | $799.96 | $799.96 |
| Speech to text: Deepgram band (Vapi only) | $95.30 | $99.40 | included | included |
| Voice: ElevenLabs band (Vapi only) | $146.00 | $238.30 | included | included |
| LLM: Vapi OpenAI band; ElevenLabs GPT-5 Mini (low) and GPT-5.5 (high) | $77.40 | $452.30 | $23.81 | $446.08 |
| Twilio outbound: 10,000 × $0.0140 | $140.00 | $140.00 | $140.00 | $140.00 |
| Twilio local number, 1 per month | $1.15 | $1.15 | $1.15 | $1.15 |
| Capacity for 10 calls | $29.00 (Core) | $29.00 (Core) | $0.00 (in Scale) | $0.00 (in Scale) |
| Monthly total | $988.85 | $1,460.15 | $964.92 | $1,387.19 |
| Effective cost per minute | $0.0989 | $0.1460 | $0.0965 | $0.1387 |
How we checked it. Once line by line, as above, and once by summing per-minute rates first. Vapi low: ($0.05 + $0.00953 + $0.0146 + $0.00774 + $0.014) × 10,000 + $1.15 + $29 = $988.85. ElevenLabs low: $299 + (10,000 − 3,738) × $0.08 + ($0.002380986 + $0.014) × 10,000 + $1.15 = $964.92. Both methods agree on all four columns.
What Scenario A says: with these two models ElevenLabs comes out $23.93 lower at the low end and $72.96 lower at the high end. Without the model line the low end reverses ($911.45 on Vapi against $941.11), so the low-end gap is entirely ElevenLabs’ lower model estimate. The swing from the low to the high model estimate is $374.90 on Vapi and $422.27 on ElevenLabs (GPT-5 Mini to GPT-5.5), several times the gap between vendors. Our Vapi vs Retell AI pricing breakdown uses the same Vapi rates and method.
Vapi vs ElevenLabs at 50,000 minutes and 30 simultaneous calls
Scenario B is the same stack at 50,000 minutes a month, a peak of 30 simultaneous calls and 10 Twilio local numbers. On Vapi, 30 lines is cheapest as Core plus 20 extra lines ($29 + $200 = $229); Pro includes 30 lines but costs its $999 minimum, because 10% of $2,500 in hosting is $250. On ElevenLabs, Scale’s 30 calls are included.
| Line item (Scenario B) | Vapi (low) | Vapi (high) | ElevenLabs (low) | ElevenLabs (high) |
|---|---|---|---|---|
| Platform: 50,000 × $0.05; Scale $299 + 46,262 × $0.080 | $2,500.00 | $2,500.00 | $3,999.96 | $3,999.96 |
| Speech to text (Vapi only) | $476.50 | $497.00 | included | included |
| Voice: ElevenLabs band (Vapi only) | $730.00 | $1,191.50 | included | included |
| LLM: Vapi OpenAI band; ElevenLabs GPT-5 Mini (low) and GPT-5.5 (high) | $387.00 | $2,261.50 | $119.05 | $2,230.40 |
| Twilio outbound: 50,000 × $0.0140 | $700.00 | $700.00 | $700.00 | $700.00 |
| Twilio local numbers, 10 per month | $11.50 | $11.50 | $11.50 | $11.50 |
| Capacity for 30 calls | $229.00 | $229.00 | $0.00 | $0.00 |
| Monthly total | $5,034.00 | $7,390.50 | $4,830.51 | $6,941.86 |
| Effective cost per minute | $0.1007 | $0.1478 | $0.0966 | $0.1388 |
With the same two models ElevenLabs is $203.49 lower at the low end and $448.64 lower at the high end. Without the model line, Vapi’s low end would be $64.46 cheaper, so again the low-end gap comes from the model estimates. The capacity line now costs Vapi $229 and ElevenLabs nothing, which is the $0.08 rule at work.
Two catches on the ElevenLabs side of Scenario B. First, ElevenLabs’ batch calling documentation says batch calls “automatically utilize the minimum of either 50% of your workspace’s concurrency limit or 70% of your agent’s concurrency limit.” On Scale that is at most 15 simultaneous batch calls, and on Business 20. To have 30 outbound calls live at once through Batch Calling you would need a workspace limit of 60, which is above Business and therefore an Enterprise conversation; the alternative is placing calls one by one through the Twilio outbound-call API from your own dialler. Second, usage beyond the plan is prepaid, and the model fees come out of the same balance. ElevenLabs’ pricing FAQ says model fees “will be deducted from your ElevenLabs credits”, and a tooltip on the same page says the displayed minute estimates are for “a custom LLM option rather than the LLM integrations we provide”, so a plan’s included minutes stretch less far once ElevenLabs bills the model. Its help centre says you need Pay As You Go credits “before you can exceed your credit quota”, and its Pay As You Go page caps top-ups at $600 a month on Scale unless two-factor authentication is on. Beyond the $299 plan, Scenario A needs $524.77 (GPT-5 Mini) to $947.04 (GPT-5.5) of minutes and model fees, so the GPT-5.5 case already breaks the cap; Scenario B needs $3,820.01 to $5,931.36. Turn on 2FA and auto top-up first: “If your balance hits $0 and Auto Top Up is off (or fails), your service will pause immediately.” Its pricing FAQ also says plans “are billed by call minutes, not by the shared credit pool”, so confirm how minutes and model fees net off before you set a top-up threshold.
Vapi’s Campaigns draw on the full subscription pool instead. Its campaigns overview says a campaign’s Max concurrency “does not reserve capacity exclusively for the campaign or increase your subscription’s total concurrency”.
Running outbound and weighing build vs buy? Zian AI runs autonomous AI sales agents across phone, SMS, email and WhatsApp, with SmartReach AI™ pacing follow-ups across channels and PrecisionPitch AI™ split-testing scripts, in 30+ languages. It is in partnership-application beta. Apply For Partnership
Twilio, SIP trunks and outbound batch calling on Vapi and ElevenLabs
Neither platform is a phone carrier for outbound calls. Both connect to one you already pay.
| Telephony question | Vapi | ElevenLabs Agents |
|---|---|---|
| Twilio | Import a Twilio number using your Twilio credentials; Twilio SIP trunk guide also published | Native integration: purchased Twilio numbers (inbound and outbound) or Verified Caller IDs (outbound only) |
| SIP trunk | Bring-your-own SIP trunk; docs index carries guides for Twilio, Telnyx, DIDWW, didlogic, Zadarma, Plivo and Amazon Chime SDK | SIP trunking for inbound and outbound; G.722 or G.711 codecs; TLS; docs index carries carrier guides for Telnyx, Plivo, Bandwidth and Sinch SIP trunks, Exotel, and Vonage (WebSocket connector), plus contact-centre integrations |
| Free numbers | 1 free US number (5 on Core, 10 on Pro); inbound only, US national only, cannot run Campaigns, cannot be ported out | Connect your own provider, per the pricing page FAQ |
| Bulk outbound tool | Campaigns: CSV up to 10,000 contacts, runs once, cannot be edited after launch; free numbers not supported | Batch Calling: CSV or XLS, schedule or send now; uses at most 50% of workspace concurrency; no Zero Retention Mode |
| When all lines are busy | “When all slots are busy, you cannot start an outbound call or accept an inbound call until a slot becomes available” | Optional call queueing for inbound Twilio and SIP calls; not supported for outbound or batch calls; queue time not billed |
| Carrier rounding (Twilio) | Twilio rounds partial minutes up to the next full minute on Programmable Voice, Twilio Client, SIP Interface, Elastic SIP Trunking and Media Streams, whichever platform sits behind it | |
Our worked tables price the carrier on exact minutes, so Twilio’s rounding makes them a floor on both platforms. If you are deciding whether you need a trunk at all, our explainer on whether an AI voice agent needs a SIP trunk walks through the cases.
For inbound, Vapi’s pricing page lists Twilio at $0.008 a minute but Twilio’s own page lists local inbound at $0.0085; budget from the carrier’s page.
Data retention, zero retention and HIPAA on each platform
| Question | Vapi | ElevenLabs Agents |
|---|---|---|
| Default retention | Raw call data kept 14 days (Usage only), 30 (Core), 180 (Pro), custom (Premier); extended retention add-on | Conversation data kept 2 years by default; set per agent in days, −1 for unlimited, 0 for scheduled deletion |
| Zero retention | Zero Data Retention, organisation-wide; pricing page lists “ZDR included” from Core ($29) up; call history, cost and latency still kept | Zero Retention Mode: the docs label it “(Enterprise)” and say it “is available to select enterprise customers”; cannot be used with batch calls. Separately, the Privacy page’s “Maximum Privacy” configuration: “Disable audio saving and set retention to 0 days for immediate deletion of data” |
| HIPAA and BAA | $2,000 a month add-on per organisation, “available with any package”; BAA emailed for signature; organisation-wide; compatible providers only | “Execution of a BAA … is only available for Enterprise tier subscriptions”; requires Zero Retention Mode; models limited to a published list while it is on |
| Interaction | “ZDR and HIPAA mode are mutually exclusive” | HIPAA requires ZRM, and ZRM excludes Batch Calling, so HIPAA outbound means calls placed individually |
The retention defaults point in opposite directions. Vapi keeps raw call data for 14 to 180 days depending on package; ElevenLabs keeps it for two years unless you shorten it. If your privacy policy promises short retention, change the ElevenLabs setting per agent on day one.
When each bill moves: a threshold table
Find the row that matches your situation. Monthly figures, US dollars, as at 4 October 2026. The concurrency rows compare only what each vendor charges for capacity: they value ElevenLabs minutes at its $0.08 overage rate and treat every other per-minute line as equal. They do not predict the total bill; Scenario A shows the speech and model lines can move the total either way.
| If your situation is… | Vapi | ElevenLabs Agents | Which charges more for the capacity (other per-minute lines aside) |
|---|---|---|---|
| Peak concurrency 1–4 | Included (Usage only) | Included on every plan | Neither |
| Peak concurrency 5–6 | $10–$20 (extra lines) | Starter $6; nothing extra if you use 75+ minutes | Vapi, at any volume |
| Peak concurrency 7–10 | $29 (Core) | Creator $22; nothing extra at 275+ minutes | Vapi, at any volume |
| Peak concurrency 11–20 | $39–$129 (Core + 1–10 lines) | Pro $99; nothing extra at 1,238+ minutes | Vapi from 750 minutes a month |
| Peak concurrency 21–30 | $139–$229 (Core + 11–20 lines) | Scale $299; nothing extra at 3,738+ minutes | Vapi from 2,000 minutes a month |
| Peak concurrency 31–40 | $239–$329 (Core + 21–30 lines) | Business $990; nothing extra at 12,375+ minutes | ElevenLabs below 8,262 minutes a month; Vapi from 9,388 |
| Peak concurrency above 40 | $10 per extra line | Burst calls at $0.160 a minute (up to 3× the limit, max 300), or Enterprise | Depends on how often you exceed 40 |
| Bulk outbound at N calls at once | Campaign uses up to your subscription’s lines | Batch Calling uses at most half the workspace limit, so you need 2N | ElevenLabs, which needs twice the plan concurrency |
| You need a HIPAA BAA | +$2,000 per organisation, any package | Enterprise only, custom price | Both; compare the quote |
| You need zero retention | ZDR from Core, $29 | Zero Retention Mode on Enterprise, not with batch calls; “Maximum Privacy” settings (audio saving off, retention 0 days) per agent | ElevenLabs for Zero Retention Mode itself, unless you are on Enterprise |
| Long silences or hold time on calls | Every minute billed; voice billed only when spoken | Silence over 10 seconds at a 95% discount | Vapi |
| You want a cloned ElevenLabs voice | Requires your own ElevenLabs API subscription | Business includes 10 Professional Voice Clones | Vapi, which adds a second subscription |
Read top to bottom, the table says: up to 30 simultaneous calls and from 2,000 minutes a month, ElevenLabs’ plans include capacity that Vapi charges $10–$229 for; between 31 and 40 calls the answer flips on volume; for bulk outbound, ElevenLabs’ 50% batch rule halves what you think you bought; and for HIPAA and Zero Retention Mode, Vapi publishes a self-serve price while ElevenLabs routes you to Enterprise. Our comparison of Zian AI and ElevenLabs Agents for outbound covers ElevenLabs’ certifications and language coverage, which this page does not.
The sentence on each vendor’s pages that cuts against this page
Against any Vapi figure on this page: Vapi’s cost documentation says the Cost value in its Performance Metrics “estimates costs for model comparison; it is not an exact quote”, and that on long calls “Actual cost is often higher” because “Growing conversation history is sent again with each request.” That method describes Vapi’s in-product estimate rather than the pricing-page band, but the warning about long calls applies to the model line either way. Vapi also says “Success Packages are currently available only to new Vapi customers. Customers on legacy pricing keep their existing billing structure and terms.”
Against “ElevenLabs gives concurrency away”: the batch rule above, plus burst pricing’s own warning that burst calls receive “lower priority for speech-to-text and text-to-speech processing” and “may experience higher latency for speech processing.”
Against “ElevenLabs is cheaper on this stack”: ElevenLabs labels its own per-model figures “Estimated costs”, and its low-end advantage in both scenarios comes entirely from the model line. Its high-end advantage depends on the model too: with GPT-6 Astra, the OpenAI model on its Models page with the highest calculator estimate, Scenario A on ElevenLabs would be $1,803.15, above the top of Vapi’s band. Vapi’s documented cost-estimating method rests on its own assumptions (5 requests per turn, 150 output tokens a minute, a 50% cache hit rate), and ElevenLabs’ calculator does not state its assumptions beside its figures, so the two model rows are not like for like. Run a test call on each and read the per-call cost before you trust either column’s LLM row.
Who Vapi and ElevenLabs Agents are each wrong for
Vapi is the wrong fit if you want one predictable meter: four of its five per-minute lines come from other providers’ price lists. It charges for every step of concurrency from the fifth call, so a team needing 20–30 simultaneous calls at 4,000 minutes a month pays $129–$229 for capacity ElevenLabs’ Scale plan includes. A cloned ElevenLabs voice also means a separate ElevenLabs API subscription.
ElevenLabs Agents is the wrong fit if you need HIPAA or Zero Retention Mode without an Enterprise contract (its “Maximum Privacy” settings delete data immediately, but Zero Retention Mode is the setting ElevenLabs describes as ensuring “no Personally Identifiable Information (PII) is logged during or stored after a call”), if you need high-concurrency bulk outbound through its own batch tool, or if you want to choose your own transcriber and speech vendor per component. It is also a poor fit at low volume on a big plan: Business at $990 is only “free” concurrency once you use 12,375 minutes.
Price stops deciding it in the middle: 10,000 to 50,000 minutes, under 30 calls at once, no regulated data. There our scenarios, with GPT-5 Mini and GPT-5.5 on ElevenLabs, put ElevenLabs $23.93 to $448.64 a month lower (2.4% to 6.1% of the Vapi figure), the low-end gaps coming entirely from the model estimates, so the bulk-dialling tool, developer experience and your carrier matter more than the rate card. If you would rather not assemble and operate a voice stack at all, Zian AI takes the other route: an operated outbound programme whose learning engine tracks around 420,000 data points across more than 10,000 leads a day.
What this page does not price
- Vapi Premier and ElevenLabs Enterprise contracts, which both vendors quote privately.
- A per-model LLM price on Vapi: its calculator gives a band per provider, not a figure per model. Models other than GPT-5 Mini and GPT-5.5 on ElevenLabs.
- Other Twilio charges such as recording, and your own carrier’s rate on a SIP trunk.
- Vapi’s extra workspaces ($20 a month) and ElevenLabs’ text messages ($0.003 each).
- Australian GST and currency conversion; ElevenLabs states “Prices exclude all taxes, levies and duties.”
Zian AI publishes no per-minute price during its partnership-application beta, so it is deliberately absent from every table above.
Frequently asked questions
Is Vapi cheaper than ElevenLabs Agents for phone calls?
No, not on our worked example: ElevenLabs Agents comes out slightly lower. For 10,000 US outbound minutes a month on Twilio, we calculate $988.85–$1,460.15 on Vapi, using its OpenAI band, and $964.92–$1,387.19 on ElevenLabs Agents, using two of the models in ElevenLabs’ own calculator, GPT-5 Mini and GPT-5.5, as at 4 October 2026. Moving from the low to the high model estimate moves each bill by $374.90–$422.27, far more than the $23.93–$72.96 gap, and the two vendors estimate model cost differently.
What does ElevenLabs Agents charge per minute for a phone call?
ElevenLabs’ agents pricing page lists $0.080 per additional call minute in its plan comparison table, with speech to text and text to speech included. LLM usage is billed on top, and the same page’s calculator estimates it per model, for example $0.00238 a minute for GPT-5 Mini; your Twilio or SIP provider bills the carrier side directly. Each plan’s monthly price is its included minutes at roughly $0.08, so the plan mainly buys concurrency.
How many calls can run at once on Vapi and ElevenLabs?
Vapi includes 4 concurrent calls on Usage only, 10 on Core and 30 on Pro, and sells extra lines at $10 a month each. ElevenLabs includes 4, 6, 10, 20, 30 and 40 across its Free to Business plans, with burst capacity up to 3 times the limit at double the rate. ElevenLabs batch calls use at most 50% of the workspace limit.
Can I use my own Twilio number or SIP trunk with both?
Yes. Vapi imports Twilio numbers and supports bring-your-own SIP trunks; its free numbers are inbound only, so outbound always needs an imported number. ElevenLabs connects Twilio natively, including verified caller IDs for outbound only, and supports SIP trunking for inbound and outbound calls. On both, the carrier bills you for the call minutes.
Which one should I use for HIPAA phone calls?
Vapi sells HIPAA as a $2,000-a-month add-on per organisation that works with any package, with a BAA emailed for signature. ElevenLabs says BAA execution is only available on Enterprise subscriptions and requires Zero Retention Mode, which cannot be used with batch calling. ElevenLabs lists Enterprise as custom pricing, so get that quote before deciding.
Can I use an ElevenLabs voice inside Vapi?
Yes. Vapi’s pricing calculator lists ElevenLabs as a voice provider at $0.0146–$0.02383 a minute as at 4 October 2026. For a custom or cloned ElevenLabs voice, Vapi’s documentation says an ElevenLabs API subscription is required, which is a second bill to add to the Vapi side.
Where every figure on this page comes from
| Figure | Who published it | Link | Date read |
|---|---|---|---|
| $0.05/min hosting; four components billed “with no markup”; $5 free credits | Vapi | vapi.ai/pricing | 4 October 2026 |
| Calculator bands: Deepgram, AssemblyAI, OpenAI STT; OpenAI, Anthropic, Google LLM; ElevenLabs, Vapi, Cartesia voice | Vapi | vapi.ai/pricing | 4 October 2026 |
| Transport: Twilio outbound $0.014, inbound $0.008, Vonage $0.00814, Telnyx $0.0055 | Vapi | vapi.ai/pricing | 4 October 2026 |
| Core $29; Pro 10% of hosting, $999 minimum; lines 4/10/30; numbers 1/5/10; retention 14/30/180 days; ZDR from Core | Vapi | vapi.ai/pricing | 4 October 2026 |
| Extra lines $10/month; workspaces $20/month; HIPAA $2,000/month per organisation | Vapi | vapi.ai/pricing | 4 October 2026 |
| “Every minute your assistants run is billed”; Success Packages for new customers only | Vapi | docs.vapi.ai/billing/pricing-and-success-packages | 4 October 2026 |
| HIPAA available with any package; extended retention add-on | Vapi | docs.vapi.ai/billing/manage-packages-and-add-ons | 4 October 2026 |
| Free numbers inbound only, US only, no Campaigns, no port-out | Vapi | docs.vapi.ai/free-telephony | 4 October 2026 |
| Campaigns: 10,000 contacts, runs once, Max concurrency shares the pool | Vapi | docs.vapi.ai/outbound-campaigns/overview | 4 October 2026 |
| Busy slots block outbound and inbound calls | Vapi | docs.vapi.ai/calls/call-concurrency | 4 October 2026 |
| 5 requests per turn, 150 output tokens/min, 50% cache; “not an exact quote”; long calls often higher | Vapi | docs.vapi.ai/…/understanding-cost | 4 October 2026 |
| ZDR and HIPAA mutually exclusive; ZDR keeps call history, cost, latency | Vapi | docs.vapi.ai/security-and-privacy/zero-data-retention | 4 October 2026 |
| HIPAA organisation-wide, BAA, compatible providers | Vapi | docs.vapi.ai/security-and-privacy/hipaa | 4 October 2026 |
| Cloned ElevenLabs voice needs an ElevenLabs API subscription | Vapi | docs.vapi.ai/customization/custom-voices/elevenlabs | 4 October 2026 |
| Plans $0/$6/$22/$99/$299/$990; included minutes; concurrency 4–40; $0.080 extra minute; $0.160 burst; $0.003 text; no telephony fee; prices exclude taxes; first-month promotions; 10 Professional Voice Clones on Business | ElevenLabs | elevenlabs.io/pricing/agents | 4 October 2026 |
| Calculator per-minute LLM estimates: GPT-5 Nano $0.0004761972, GPT-6 Astra $0.086204025, GPT-5 Mini $0.002380986, GPT-5.5 $0.044607915, GPT-4.1 $0.016638444, Claude Sonnet 4.5 $0.028447328, Gemini 2.5 Flash $0.002887301; “Estimated costs in USD…”; “LLM at provider cost”; model fees “will be deducted from your ElevenLabs credits”; minute estimates assume “a custom LLM option”; plans “billed by call minutes, not by the shared credit pool” | ElevenLabs | elevenlabs.io/pricing/agents | 4 October 2026 |
| “Maximum Privacy”: audio saving off, retention 0 days | ElevenLabs | elevenlabs.io/docs/eleven-agents/customization/privacy | 4 October 2026 |
| Zero Retention Mode: no PII “logged during or stored after a call” | ElevenLabs | elevenlabs.io/docs/…/privacy/zrm | 4 October 2026 |
| Silence over 10 seconds at a 95% discount; Pay As You Go credits needed to exceed quota | ElevenLabs | elevenlabs.io/docs/…/how-much-does-eleven-agents-cost | 4 October 2026 |
| Supported models, including GPT-5 Nano and GPT-6 Astra; LLM pass-through with no markup; Gemini and Claude at Vertex regional rates, 10% increase | ElevenLabs | elevenlabs.io/docs/eleven-agents/customization/llm | 4 October 2026 |
| Burst up to 3× limit or 300; 2× rate; deprioritised | ElevenLabs | elevenlabs.io/docs/eleven-agents/guides/burst-pricing | 4 October 2026 |
| Batch calls use min of 50% workspace or 70% agent concurrency; no ZRM for batch | ElevenLabs | elevenlabs.io/docs/eleven-agents/phone-numbers/batch-calls | 4 October 2026 |
| Call queueing: inbound only; queue time not billed | ElevenLabs | elevenlabs.io/docs/eleven-agents/guides/call-queueing | 4 October 2026 |
| Twilio purchased numbers and Verified Caller IDs | ElevenLabs | elevenlabs.io/docs/…/native-integration | 4 October 2026 |
| SIP trunking inbound and outbound; G.722/G.711; TLS | ElevenLabs | elevenlabs.io/docs/eleven-agents/phone-numbers/sip-trunking | 4 October 2026 |
| Retention 2 years default; −1 and 0 settings | ElevenLabs | elevenlabs.io/docs/…/privacy/retention | 4 October 2026 |
| ZRM “available to select enterprise customers” | ElevenLabs | elevenlabs.io/docs/eleven-api/resources/zero-retention-mode | 4 October 2026 |
| BAA Enterprise only; ZRM required; LLM list | ElevenLabs | elevenlabs.io/docs/eleven-agents/legal/hipaa | 4 October 2026 |
| Pay As You Go cap without 2FA: $600 on Scale; service pauses at $0 | ElevenLabs | elevenlabs.io/docs/overview/administration/pay-as-you-go | 4 October 2026 |
| US local outbound $0.0140/min; local inbound $0.0085/min; local number $1.15/month | Twilio | twilio.com/en-us/voice/pricing/us | 4 October 2026 |
| Partial minutes rounded up to the next full minute on five named products | Twilio | support.twilio.com article 223132307 | 4 October 2026 |
| Scenario totals, per-minute figures, gaps, $0.08 rule ratios, threshold crossovers, top-up amounts | Zian AI calculation from the rows above | This page | 4 October 2026 |
Still deciding whether to build at all? Zian AI’s autonomous sales agents handle calls, SMS, email and WhatsApp follow-up with CRM integrations for HubSpot, Salesforce and HighLevel, and Zian is taking partners during its partnership-application beta. Apply For Partnership