Quick answer: For 1,000 three-minute US outbound calls, we calculate $290.25 on ElevenLabs Agents (Pro plan on monthly billing, GPT-5 Mini, your own Twilio number). On Retell AI the same calls cost $296.00 with a $0.015 voice or $371.00 with an ElevenLabs Flash voice, as at 4 November 2026. Retell bills by the second and includes 20 concurrent calls. ElevenLabs bundles the voice into a $0.08 minute.
“Retell vs ElevenLabs: which should I build my phone agent on?” This page compares Retell AI (retellai.com) with ElevenLabs Agents, which ElevenLabs brands ElevenAgents. On both you write the prompt, pick a model from OpenAI, Anthropic or Google, connect Twilio or a SIP trunk, and pay per minute. Every figure was read on Retell’s, ElevenLabs’ and Twilio’s own pages on 4 November 2026, in US dollars, excluding tax such as Australian GST.
Where Zian fits if you would rather not build
If you want an AI receptionist and sales agent that is already built, Zian is the platform pick and LeadsNow.ai is the done-for-you pick. Retell and ElevenLabs sell the parts of a phone agent by the minute; you still write the prompts, wire the CRM and run the dialling. Zian is an AI sales and reception agents platform. It answers inbound calls 24/7 as an AI receptionist (qualifying, booking, transferring, taking messages), runs outbound follow-up by voice, SMS, email, WhatsApp, Instagram DMs and Facebook Messenger, and handles answering machines and voicemail.
SmartReach AI™ sets message, channel, timing and follow-up pacing; PrecisionPitch AI™ keeps split-testing scripts against real outcomes. Zian works in 30+ languages with voice cloning, runs knowledge-base lookups, connects to HubSpot, Salesforce, HighLevel and Zapier, and its live counter shows 50,769+ qualified sales appointments set. It is in a partnership-application beta with no public price; each of its three listed plans, including Intel + Widgets (Free Forever), starts with a partnership application. LeadsNow.ai is the managed option: AI appointment setting and lead generation that runs on Zian, sold pay-per-result, with leads answered in under 60 seconds, 24/7.
| Route | What you get | Who builds the agent | Published price, as at 4 November 2026 |
|---|---|---|---|
| 1. Zian (platform pick) | Ready-made AI receptionist and outbound sales agents across calls, SMS, email, WhatsApp, Instagram and Messenger | Zian’s digital-team agents; voice, rules and knowledge-base scope are configurable | No public price during the partnership-application beta |
| 2. LeadsNow.ai (done-for-you pick) | Managed AI appointment setting and lead generation that runs on Zian | LeadsNow.ai runs it | Pay-per-result; we do not print its price |
| 3. Retell AI | Voice and chat agent platform, Retell numbers, Dialer campaigns | Your team | $0.055 a minute Voice Infra plus model, voice and telephony |
| 4. ElevenLabs Agents | Voice and chat agent platform with ElevenLabs speech bundled in | Your team | Plans from $0 to $990 a month; $0.080 per additional minute plus model |
Method and disclosure: zian.ai publishes this page and appears in it. By editorial policy Zian is listed first as the platform pick and LeadsNow.ai as the done-for-you pick. Retell and ElevenLabs are compared only on facts read on their own pages on 4 November 2026. ElevenLabs also sells a ready-made receptionist, Reception.ai, for small businesses ($29 a month on monthly billing, per its own plans page); this page compares the two build platforms.
Want the receptionist and outbound follow-up without building either? Apply For Partnership
How Retell and ElevenLabs Agents build a per-minute price
Retell itemises the minute; ElevenLabs bundles most of it. Retell lists Voice Infra at $0.055 a minute, then a per-minute model, voice, telephony and add-ons, with no separate speech-to-text line. ElevenLabs’ agents pricing page lists $0.080 per additional call minute, with text to speech and speech to text included. Its FAQ says “LLM usage is billed separately on top, based on the model you choose” and that “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.”
| Component | Retell AI (pay as you go) | ElevenLabs Agents |
|---|---|---|
| Platform per minute | Retell Voice Infra $0.055 | $0.080 per minute beyond the plan’s included minutes |
| Speech to text | No separate line | Included in the call minute |
| Voice | $0.015 (Retell, Minimax, Fish, Cartesia, OpenAI, Inworld); ElevenLabs Flash $0.040, Multilingual v2 $0.06, v3 Conversational and v4 Turbo $0.08, v3 and v4 $0.10 | Included in the call minute |
| Language model | Per-minute list price, from $0.0016 (GPT 5 nano) to $0.32 (GPT 6 Astra) on the standard tier, or $0.384 for GPT 6.1 Sol (Ultrafast); GPT 5 mini $0.008; GPT 5.5 $0.16 | “At cost”, varies by model; calculator estimates per minute, for example GPT-5 Mini $0.00238 and GPT-5.5 $0.04461 |
| Knowledge base | +$0.005 a minute on calls with a knowledge base; 10 knowledge bases free, then $8 a month each | Listed as an included feature |
| Telephony | $0.015 a minute on Retell’s US Twilio numbers; no charge for SIP trunking or custom telephony | No ElevenLabs fee; your carrier bills you (Twilio US local outbound $0.0140 a minute) |
| Phone number | Retell number $2.00 a month (US or Canada); toll-free $5 | Your carrier’s number (Twilio local $1.15 a month) |
| Plan fee | None; prepaid credits | Free $0, Starter $6, Creator $22, Pro $99, Scale $299, Business $990; Enterprise custom |
| Free start | $10 in trial credits, once per email address | Free plan with 15 minutes of calls |
The table hides one asymmetry. Retell’s model prices are list prices per minute. ElevenLabs’ are estimates: its calculator footnote reads “Estimated costs in USD, covering ElevenAgents platform voice and LLM models.” On GPT-5 Mini, Retell lists $0.008 a minute and ElevenLabs estimates $0.00238. On GPT-5.5, Retell lists $0.16 and ElevenLabs estimates $0.04461. So the model line is a list price against an estimate, not two invoices.
What 1,000 three-minute calls cost on Retell vs ElevenLabs
Worked calculation: every line is a published rate times a quantity you control. Inputs: 1,000 connected US-to-US outbound calls in a month, each exactly 3 minutes 0 seconds (3,000 minutes), GPT-5 Mini and a knowledge base on every call, one phone number, no more than 20 calls at once, no long silences, prompts under 4,000 tokens, no Enterprise contract.
- Retell A uses an ElevenLabs Flash voice through Retell ($0.040) and Retell’s US Twilio numbers ($0.015). The rate is $0.055 + $0.008 + $0.040 + $0.005 + $0.015 = $0.123 a minute.
- Retell B is the same, with a $0.015 Retell Platform voice: $0.098 a minute.
- ElevenLabs runs on the Pro plan ($99 on monthly billing, 1,238 minutes, 20 concurrent calls), chosen to match Retell’s 20 included calls. Extra minutes cost $0.080, and the model is ElevenLabs’ GPT-5 Mini estimate. The carrier is your own Twilio local number at $0.0140 a minute outbound plus $1.15 a month.
| Line (1,000 calls × 3 minutes) | Retell A (ElevenLabs Flash voice) | Retell B ($0.015 voice) | ElevenLabs Agents (Pro) |
|---|---|---|---|
| Platform | $165.00 (3,000 × $0.055) | $165.00 | $239.96 ($99 + 1,762 × $0.080) |
| Model (GPT-5 Mini) | $24.00 (3,000 × $0.008) | $24.00 | $7.14 (3,000 × $0.00238, estimate) |
| Voice | $120.00 (3,000 × $0.040) | $45.00 (3,000 × $0.015) | Included |
| Knowledge base | $15.00 (3,000 × $0.005) | $15.00 | Included |
| Telephony | $45.00 (3,000 × $0.015) | $45.00 | $42.00 (3,000 × $0.0140, Twilio) |
| Phone number, one month | $2.00 | $2.00 | $1.15 |
| Concurrency (20 or fewer) | $0.00 | $0.00 | $0.00 (in Pro) |
| Total for 1,000 calls | $371.00 | $296.00 | $290.25 |
| Cost per call | $0.371 | $0.296 | $0.290 |
| Same calls with GPT-5.5 | $827.00 | $752.00 | $416.93 |
| 10,000 calls (30,000 minutes, 5 numbers) | $3,700.00 | $2,950.00 | $2,897.14 |
Checked a second way: Retell A is $0.123 × 3,000 + $2 = $371.00; ElevenLabs is $99 + (3,000 − 1,238) × $0.08 + 3,000 × ($0.002380986 + $0.014) + $1.15 = $290.25. The GPT-5.5 row swaps only the model line ($0.16 on Retell, $0.044607915 on ElevenLabs). The 10,000-call row keeps Pro, because at 30,000 minutes every ElevenLabs plan from Starter to Business costs within 4 cents of $2,400 before model and carrier.
Two lines outside the ledger are each bigger than the $5.75 gap between ElevenLabs and Retell B. On yearly billing ElevenLabs lists Pro at $82.50 a month (“2 months free”), which takes $16.50 off and puts ElevenLabs at $273.75. And Twilio’s US voice pricing page lists Media Streams at $0.0044 a minute: ElevenLabs’ register-call guide says it connects Twilio calls to the agent “via WebSocket”, and its native-integration page does not say whether that leg bills as Media Streams. If your Twilio invoice shows it, add $13.20 per 1,000 three-minute calls, which puts ElevenLabs at $303.45, above Retell B. Read Retell B against ElevenLabs as a near tie that your billing term and your Twilio invoice decide; Retell A against ElevenLabs is not close.
The ledger shows three things. ElevenLabs comes out $5.75 below Retell B on GPT-5 Mini, and that gap is smaller than the $16.86 model-line difference between the two vendors’ figures. An ElevenLabs voice bought through Retell moves Retell up $75 per 1,000 calls. And on a premium model the gap widens to $335.07 or more, because Retell’s list price for GPT-5.5 is about 3.6 times ElevenLabs’ estimate. If you are pricing Retell alone at other volumes, our Retell AI cost breakdown at 1,000 to 20,000 minutes works through three more stacks.
The voice-line test: which platform the voice you want makes cheaper
The voice-line test: before comparing the two, add Retell’s $0.055 Voice Infra to the price of the voice you want, and set the total against ElevenLabs’ $0.080 bundled minute. With any ElevenLabs voice through Retell, Voice Infra plus voice comes to $0.095 (Flash) to $0.155 (v3 or v4) a minute, above $0.080 on every ElevenLabs model Retell lists. With a $0.015 voice from Retell, Minimax, Fish, Cartesia, OpenAI or Inworld it comes to $0.070, below $0.080.
So the voice decides the first cut. Want an ElevenLabs voice? The bundled minute starts $0.015 to $0.075 cheaper. Happy with a $0.015 voice? Retell’s speech side starts $0.010 cheaper, and the model line decides the rest: a list price on Retell, an estimate deducted from credits on ElevenLabs. Run a test call on each and read the per-call cost.
How each one rounds a call: seconds, minutes and the carrier
Retell’s pricing FAQ says: “Each call is tracked to the nearest second. At the end of each billing cycle, we charge based on your total accumulated minutes. No rounding up per call, no inflated bills.” ElevenLabs’ cost help article says voice calls “are charged based on the call duration, with a 95% discount for periods of silence longer than 10 seconds,” and that “the length of the call is measured based on the connection duration,” which runs “from when you begin the call, to when you end the call or the window is closed. This is why the call itself may be shorter than the duration you are charged for.” We could not find a stated rounding increment for ElevenLabs call minutes on its agents pricing page, its cost help article, its Pay As You Go page or its billing page, as at 4 November 2026. Its 26 October 2026 changelog adds that “teardown time after the call is no longer billed.”
The carrier brings whole minutes back. Twilio says “any partial minutes (under 60 seconds) are rounded up to the next full minute” on Programmable Voice, Twilio Client, SIP Interface, Elastic SIP Trunking and Media Streams. On ElevenLabs your Twilio number bills that way; on Retell, only if you bring Twilio over your own SIP trunk.
| Same 1,000 calls, each 2 minutes 20 seconds | Retell A | Retell B | ElevenLabs Agents (Pro) |
|---|---|---|---|
| Billed platform minutes | 2,333.33 (per second) | 2,333.33 | 2,333.33 (assumes billed connection time equals talk time, with no rounding) |
| Billed carrier minutes | 2,333.33 (Retell numbers) | 2,333.33 | 3,000 (Twilio rounds each call up to 3) |
| Total | $289.00 | $230.67 | $235.33 |
At 3 minutes 0 seconds, ElevenLabs is $5.75 below Retell B. At 2 minutes 20 seconds, Retell B is $4.66 below ElevenLabs on monthly billing, because Twilio bills 3,000 carrier minutes for 2,333 minutes of talk; on yearly billing ElevenLabs comes back to $218.83, and a Twilio Media Streams charge would add $13.20 to it. Retell has rules of its own: calls under 10 seconds with a dynamic opening message bill as 10 seconds, and above 4,000 prompt tokens “Billed Duration = Original Duration × Scaling Factor (rounded up)”.
How many concurrent calls Retell and ElevenLabs include
Retell gives every pay-as-you-go workspace 20 concurrent calls and sells more at $8 per concurrent call per month. ElevenLabs ties concurrency to the plan: 4 on Free, 6 on Starter, 10 on Creator, 20 on Pro, 30 on Scale and 40 on Business, with “Elevated concurrency limits” on Enterprise. Each plan’s price is roughly its included minutes at $0.08, so once you use them the step up mainly buys concurrency.
| Peak simultaneous calls | Retell AI, extra per month | ElevenLabs Agents plan |
|---|---|---|
| Up to 20 | $0 (20 included) | Pro $99 (1,238 minutes included); lower plans cover 4 to 10 |
| 21 to 30 | $8 to $80 | Scale $299 (3,738 minutes included) |
| 31 to 40 | $88 to $160 | Business $990 (12,375 minutes included) |
| Above 40 | $8 per extra call; Enterprise “starting at 50+ calls” | Enterprise, or burst |
| Burst above your limit | +$0.10 a minute on the whole call; up to the lower of 3× your limit or your limit plus 300 | $0.160 a minute (double rate); up to 3× your limit or 300, whichever is lower, for non-enterprise |
| When every slot is busy | Inbound waits about 40 seconds, then goes to a fallback number or ends | Optional call queueing for inbound Twilio and SIP calls; queue time not billed; not for outbound or batch calls |
At 3,000 minutes and 40 calls at once, Retell adds $160 to the ledger; ElevenLabs means Business at $990 against Pro’s $239.96, a $750.04 step that shrinks to roughly nothing from 12,375 minutes. ElevenLabs’ burst guide also warns that burst calls get “lower priority for speech-to-text and text-to-speech processing.”
Building a calling programme rather than buying one? Zian’s Outbound Appointment Setter and AI Receptionist run on one platform, with SmartReach AI™ pacing the follow-up. Apply For Partnership
Own numbers, SIP trunks and Twilio on each platform
Retell sells numbers; ElevenLabs connects the ones you have. Retell sells US and Canada numbers and calls 15 countries (Australia $0.10 a minute on its Twilio route); elsewhere you bring a carrier. ElevenLabs’ pricing FAQ says you “connect your own provider.” We found no page in ElevenLabs’ agents documentation index offering to sell a number for ElevenAgents, as at 4 November 2026.
| Telephony question | Retell AI | ElevenLabs Agents |
|---|---|---|
| Buy a number from the platform | Yes: US or Canada, $2 a month; US toll-free $5 | Connect your own provider |
| Twilio | Retell numbers run on Twilio or Telnyx; your own Twilio number connects over elastic SIP trunking | Native integration: purchased Twilio numbers (inbound and outbound) or Verified Caller IDs (outbound only) |
| SIP trunking | Elastic SIP trunking or dial-to-SIP-URI; TCP, UDP, TLS and mTLS; guides for Twilio, Telnyx, Vonage, Avaya, Genesys, Five9 and Amazon Connect | Inbound and outbound SIP trunking over TCP or TLS (UDP for testing only); guides for Telnyx, Plivo, Exotel, Bandwidth, Sinch and Vobiz, a Vonage WebSocket connector, plus Amazon Connect, Genesys and Five9 |
| Platform fee on your own carrier | “No Charge for Sip Trunking/Custom Telephony” | No ElevenLabs telephony fee |
| Carrier rounding | Per second on Retell numbers; your carrier’s rules on your own trunk | Your carrier’s rules (Twilio: whole minutes, rounded up) |
For inbound reception lines the carrier rate changes on ElevenLabs. Twilio lists US local inbound at $0.0085 a minute. Retell prints one $0.015 rate for its US Twilio route. If you are weighing a trunk at all, the Vapi vs ElevenLabs Agents phone-call comparison covers ElevenLabs’ Twilio and SIP set-up against a third platform.
Batch and outbound calling: Retell Dialer vs ElevenLabs Batch Calling
Retell has moved bulk outbound to its Dialer: “For new outbound campaigns, use the Dialer, which adds retries, SMS steps, and compliance checks. Batch calls are still available.” Its pricing page still lists Batch Call at +$0.005 a dial. Dialer campaigns run call and SMS sequences in each contact’s calling window, screen against federal and state Do Not Call registries (run under your own FTC registry account), TCPA litigators and line types (billed per number checked; only Retell’s internal DNC check is on by default), and dial from local area codes with up to 500 numbers per campaign. We could not find a per-dial fee for Dialer campaigns on Retell’s pricing page or the Dialer pages we read, as at 4 November 2026; those pages do price numbers, concurrency and daily spam checks ($2 per number per month).
ElevenLabs Batch Calling takes a CSV or XLS list, sends now or on a schedule, and works with native Twilio and SIP trunk numbers; its API includes a retry-batch-job endpoint. Its changelog added Twilio answering-machine detection to batch calls on 7 September 2026, and hang-up actions on 19 October. The concurrency rules differ:
- ElevenLabs: batch calls automatically use the lower of 50% of the workspace’s concurrency limit or 70% of the agent’s limit, so Pro’s 20 calls give at most 10 batch calls at once.
- Retell: a Dialer campaign’s limit defaults to half the workspace limit in the dashboard, and you can change it. Dialer calls “never use burst; they wait for a free slot instead.”
- Zero retention: ElevenLabs says Zero Retention Mode “cannot be enabled for batch calls.”
Knowledge base limits and what each charges for it
Retell charges for knowledge-base use and caps sources; ElevenLabs includes it and caps size. Retell bills $0.005 per minute of calls with a knowledge base enabled ($15.00 on our ledger), with 10 knowledge bases free and $8 a month each after that; each holds up to 500 URLs, 50 text snippets and 25 documents of up to 50 MB. ElevenLabs lists knowledge bases and RAG as included, takes files up to 20MB and caps the total indexed for RAG by tier: 1MB Free, 2MB Starter, 20MB Creator, 100MB Pro, 500MB Scale, 1GB Business. It says RAG adds “around 250ms” of latency.
Latency: what Retell and ElevenLabs each claim, and what it measures
These are vendor-stated figures that measure different things. Retell’s home page claims “~600ms latency”, which its documentation defines as “measured from when the user stops speaking to when the AI agent begins responding.” ElevenLabs’ agents page promises “ultra-low latency” with no end-to-end number; its methodology page gives server-side component medians (about 100ms for Eleven v4 Turbo and Flash v2.5, about 150ms to a committed transcript for Scribe v2 Realtime) that “exclude network round-trips”. So do not set 600ms against 100ms: measure end-to-end on your own calls on each platform. Our explainer on what vendor voice-agent latency numbers measure sets out how.
HIPAA, SOC 2 and BAAs as each vendor states them
Retell offers a self-signed BAA at no extra fee; ElevenLabs, on Enterprise only. Retell’s security and compliance page says it is “HIPAA compliant, GDPR compliant, SOC 2 Type 1 & Type 2 certified”. It says its BAA and DPA are “available for self-signing” and that “There is no additional fee to sign these agreements.” Its home page adds ISO 27001 (“Fully compliant with HIPAA, SOC2 Type II, GDPR and ISO 27001”), which the compliance documentation does not list, and its pricing page says “SOC 2 certified, HIPAA-ready.” The compliance page also says Retell does “not currently operate services within the European Union.”
ElevenLabs’ agents page says “We are SOC 2 Type II, ISO 27001, and PCI DSS Level 1 certified, with HIPAA and GDPR compliance”, and badges ISO 42001 as well. It lists data residency across the US, EU and India among its Enterprise controls. Its HIPAA page says “Execution of a BAA … is only available for Enterprise tier subscriptions.” HIPAA use also requires Zero Retention Mode, which cannot run with batch calls, so HIPAA outbound on ElevenLabs means placing calls one by one.
Retell AI vs ElevenLabs Agents: which one to choose
| If your situation is… | Lean towards | Why, from the figures above |
|---|---|---|
| You want an ElevenLabs voice | ElevenLabs Agents | Voice Infra plus an ElevenLabs voice on Retell is $0.095 to $0.155 a minute against the $0.080 bundled minute |
| A $0.015 voice is fine and calls are short or uneven | Retell AI | Per-second billing on Retell numbers; $230.67 against $235.33 per 1,000 calls at 2 minutes 20 seconds with ElevenLabs Pro on monthly billing (yearly billing brings ElevenLabs to $218.83) |
| You need 21 to 40 calls at once on low volume | Retell AI | $8 to $160 a month extra, against moving to Scale ($299) or Business ($990) |
| You need a HIPAA BAA without an Enterprise contract | Retell AI | Self-signed BAA, no extra fee; ElevenLabs BAAs are Enterprise only |
| You want DNC and TCPA-litigator screening inside the campaign tool | Retell AI | Dialer compliance checks, billed per number; we could not find DNC screening in ElevenLabs’ batch-calling docs or docs index |
| You need EU or India data residency, or ISO 42001 or PCI DSS on paper | ElevenLabs Agents | Stated on its agents page (residency is listed among Enterprise controls); Retell says it does not operate services in the EU, though its home page offers Enterprise on-prem deployment “to meet strict security and data residency requirements” |
| You want to buy numbers from the platform | Retell AI | US and Canada numbers at $2 a month; ElevenLabs connects your own provider |
| You would rather not build at all | Zian, or LeadsNow.ai done-for-you | Ready-made receptionist and outbound agents; managed, pay-per-result service |
The voice and the call length decide cost by cents per call; concurrency, HIPAA paperwork and the bulk-dialling tool decide most real choices. If you are still weighing build against buy, our guide to how to build an AI receptionist, and when not to covers the work either platform leaves with you.
The sentence on each vendor’s pages that cuts against this page
Against “ElevenLabs is cheaper on our ledger”: its model figures are labelled “Estimated costs”. A tooltip on the same page says the displayed minute estimates are for “a custom LLM option rather than the LLM integrations we provide”, and model fees are “deducted from your ElevenLabs credits.” Without 2FA, top-ups are capped at $200 a month on Pro, and the GPT-5.5 case needs $274.78 (the 10,000-call row about $2,372); “If your balance hits $0 and Auto Top Up is off (or fails), your service will pause immediately.”
Against “Retell is cheaper on short calls”: Retell bills silence and hold time, and the 4,000-token rule counts the transcript and retrieved knowledge-base content. Retell pay as you go is prepaid, and legacy monthly-billing workspaces moved to credits on 30 September 2026: “When your credit balance reaches zero, calls, chat, and other usage stop,” and “Credits never expire, but they are non-refundable once purchased.”
What this page does not price
Enterprise contracts on either platform; Retell’s other add-ons (Advanced Denoising and Safety Guardrails +$0.005 a minute, PII Removal +$0.01, AI Quality Assurance, Branded Call +$0.10 per outbound call, verified numbers $10 a month) and speech-to-speech models; ElevenLabs text messages ($0.003), its 95% silence discount and first-month promotions; Twilio extras such as recording ($0.0025 a minute); and Fast Tier models on either platform. Zian and LeadsNow.ai publish no per-minute price, so neither appears in the cost tables.
Frequently asked questions
Is Retell AI cheaper than ElevenLabs Agents?
Not on our three-minute ledger, and only narrowly. 1,000 three-minute US outbound calls on GPT-5 Mini cost $290.25 on ElevenLabs Pro (monthly billing) with your own Twilio number. They cost $296.00 on Retell with a $0.015 voice and $371.00 with an ElevenLabs Flash voice, as at 4 November 2026. At 2 minutes 20 seconds a call, Retell with a $0.015 voice drops to $230.67 against $235.33, because Twilio rounds each call up to whole minutes. Yearly billing on ElevenLabs, or a Twilio Media Streams charge, moves either total by more than these gaps.
Can I use an ElevenLabs voice on Retell?
Yes. Retell’s pricing page lists ElevenLabs voices per model: Flash $0.040 a minute, Multilingual v2 $0.06, v3 Conversational and v4 Turbo $0.08, and v3 and v4 $0.10. That is on top of Retell’s $0.055 Voice Infra, so an ElevenLabs voice through Retell costs $0.095 to $0.155 a minute before the model and telephony.
How many concurrent calls do Retell and ElevenLabs include?
Retell includes 20 concurrent calls on pay as you go and charges $8 a month for each extra one. ElevenLabs includes 4, 6, 10, 20, 30 and 40 on its Free, Starter, Creator, Pro, Scale and Business plans. Both offer burst above the limit: Retell adds $0.10 a minute to the whole call, and ElevenLabs charges $0.160 a minute.
Does Retell or ElevenLabs bill by the second?
Retell says each call is tracked to the nearest second with no rounding up per call. ElevenLabs says calls are charged on call duration, measured as connection duration; we could not find a stated rounding increment as at 4 November 2026. If your ElevenLabs calls run on Twilio, Twilio’s billing article says partial minutes are rounded up to the next full minute.
Which one signs a HIPAA BAA?
Retell’s compliance page says its BAA is available for self-signing with no additional fee. ElevenLabs’ HIPAA page says BAA execution is only available on Enterprise subscriptions. ElevenLabs also requires Zero Retention Mode for HIPAA, and that mode cannot be used with batch calls.
Can I bring my own Twilio number or SIP trunk?
Yes, on both. Retell connects your Twilio number over elastic SIP trunking, charges nothing for custom telephony, and sells US and Canada numbers at $2 a month. ElevenLabs connects Twilio natively, including Verified Caller IDs for outbound only, and supports inbound and outbound SIP trunking. Your carrier bills the call minutes.
Where every figure on this page comes from
| Figure | Who published it | Link | Date read |
|---|---|---|---|
| Voice Infra $0.055/min; voice prices ($0.015; ElevenLabs $0.040 to $0.10); LLM per-minute prices (GPT 5 nano $0.0016, GPT 5 mini $0.008, GPT 5.5 $0.16, GPT 6 Astra $0.32); US Twilio telephony $0.015; no charge for SIP trunking/custom telephony; Knowledge Base +$0.005/min; Batch Call +$0.005/dial; other add-ons; $2 numbers; $8 concurrency; 10 free knowledge bases; $10 trial credits; speech-to-speech $0.07 to $0.38 | Retell AI | retellai.com/pricing | 4 November 2026 |
| Billed to the nearest second, no rounding up per call; silence billed; 20 concurrent calls; Enterprise “starting at 50+ calls”; “SOC 2 certified, HIPAA-ready” | Retell AI | retellai.com/pricing (FAQ) | 4 November 2026 |
| “~600ms latency”; “HIPAA, SOC2 Type II, GDPR and ISO 27001”; Enterprise on-prem “to meet strict security and data residency requirements” | Retell AI | retellai.com | 4 November 2026 |
| 600ms measured from user stop to agent response | Retell AI | Check estimated latency | 4 November 2026 |
| Prepaid credits; usage stops at zero; credits non-refundable; $10 trial once per email; legacy billing moved 30 September 2026 | Retell AI | Billing: credits, auto recharge, and invoices | 4 November 2026 |
| 10-second minimum; 4,000-token scaling formula | Retell AI | Exceptions to Our Per-Minute Pricing | 4 November 2026 |
| 20 calls per workspace; burst +$0.10/min, lower of 3× or limit plus 300; inbound wait about 40 seconds; Dialer never bursts | Retell AI | Understand concurrency & limits | 4 November 2026 |
| Batch calls deprecated, still available | Retell AI | Create and schedule batch calls | 4 November 2026 |
| Dialer sequences, retries, SMS steps, DNC/TCPA litigator screening | Retell AI | Outbound call campaigns; Compliance checks | 4 November 2026 |
| Campaign limit defaults to half the workspace limit; 500 numbers per campaign; $2 local, $5 toll-free; $2 daily spam check; $8 concurrency | Retell AI | Caller ID and pacing | 4 November 2026 |
| US and Canada numbers; calls to 15 countries | Retell AI | Purchase phone number | 4 November 2026 |
| Australia $0.10/min on Retell’s Twilio route | Retell AI | International calling | 4 November 2026 |
| Elastic SIP trunking, dial-to-SIP-URI; TCP, UDP, TLS, mTLS | Retell AI | Custom telephony; Twilio | 4 November 2026 |
| Knowledge base $0.005/min, 10 free then $8; 500 URLs, 50 snippets, 25 documents, 50 MB | Retell AI | Knowledge base setup | 4 November 2026 |
| HIPAA, GDPR, SOC 2 Type 1 and Type 2; self-signed BAA and DPA, no additional fee; no services in the EU | Retell AI | Security and compliance | 4 November 2026 |
| Plans $0/$6/$22/$99/$299/$990; included minutes 15 to 12,375; concurrency 4 to 40; $0.080 extra minute; $0.160 burst; $0.003 text; no telephony fee; LLM “at cost”; per-model LLM estimates (GPT-5 Mini $0.002380986, GPT-5.5 $0.044607915); “Estimated costs in USD…”; custom-LLM tooltip; first-month $1 and $11; Pro $82.50 a month on yearly billing; Enterprise BAAs | ElevenLabs | elevenlabs.io/pricing/agents | 4 November 2026 |
| SOC 2 Type II, ISO 27001, PCI DSS Level 1, HIPAA and GDPR; ISO 42001 badge; US, EU and India data residency; “ultra-low latency” | ElevenLabs | elevenlabs.io/agents | 4 November 2026 |
| Charged on call duration, connection duration (“may be shorter than the duration you are charged for”); 95% silence discount over 10 seconds | ElevenLabs | How much does ElevenAgents cost? | 4 November 2026 |
| Teardown no longer billed (26 October 2026); batch answering-machine detection (7 September, 19 October 2026) | ElevenLabs | ElevenLabs changelog | 4 November 2026 |
| Burst up to 3× or 300; 2× rate; deprioritised processing | ElevenLabs | Burst pricing | 4 November 2026 |
| Call queueing inbound only; queue time not billed | ElevenLabs | Call queueing | 4 November 2026 |
| Batch calls use min of 50% workspace or 70% agent concurrency; no ZRM for batch | ElevenLabs | Batch calling | 4 November 2026 |
| Register-call route returns TwiML that connects the call “via WebSocket” | ElevenLabs | Register Twilio calls | 4 November 2026 |
| Twilio purchased numbers and Verified Caller IDs | ElevenLabs | Twilio native integration | 4 November 2026 |
| SIP trunking inbound and outbound, TLS | ElevenLabs | SIP trunking | 4 November 2026 |
| 20MB files; RAG limits 1MB to 1GB by tier; RAG about 250ms | ElevenLabs | Knowledge base; RAG | 4 November 2026 |
| Server-side medians: v4 Turbo and Flash v2.5 about 100ms, v3 Conversational about 165ms, Scribe v2 Realtime about 150ms; not comparable with client figures | ElevenLabs | How ElevenLabs latency figures are measured | 4 November 2026 |
| BAA Enterprise only; Zero Retention Mode required; model list | ElevenLabs | HIPAA | 4 November 2026 |
| Top-up cap without 2FA ($200 on Pro); service pauses at $0 | ElevenLabs | Pay As You Go | 4 November 2026 |
| Reception.ai from $29 a month | ElevenLabs | Reception.ai plans and pricing | 4 November 2026 |
| US local outbound $0.0140/min; local inbound $0.0085/min; local number $1.15/month; recording $0.0025/min; Media Streams $0.0044/min | Twilio | twilio.com/en-us/voice/pricing/us | 4 November 2026 |
| Partial minutes rounded up to the next full minute on five named products | Twilio | Twilio support article 223132307 | 4 November 2026 |
| Ledger totals, per-call costs, 2 minute 20 second case, voice-line test thresholds, concurrency steps | Zian AI calculation from the rows above | This page | 4 November 2026 |
Rather run calls than build the stack? Zian answers inbound calls as an AI receptionist and books from outbound follow-up, and is taking partners during its partnership-application beta. Apply For Partnership