Vapi vs Twilio: Build Direct or Add Vapi on Top? - Zian AI

Vapi vs Twilio: Build Direct or Add Vapi on Top?

Quick answer: Vapi and Twilio are different products. Twilio carries the call and sells the plumbing to an agent (ConversationRelay at $0.07 a minute, Media Streams at $0.0044); Vapi runs the agent for $0.05 a minute plus models, and needs a carrier such as Twilio to dial out. At 10,000 inbound US minutes, four routes cost $448.85–$1,376.15 before your own servers.

The question “Vapi vs Twilio” usually comes from a team that has a voice-agent prototype and has to decide what it will run on. One option is to build straight on Twilio Programmable Voice, either with ConversationRelay (Twilio runs speech-to-text and text-to-speech, your server trades text with a large language model) or with bidirectional Media Streams (raw call audio over a WebSocket, and you bring Deepgram, OpenAI, ElevenLabs or whichever STT, LLM and TTS vendors you like). The other is to put Vapi on top: Vapi orchestrates the transcriber, model and voice, and the call arrives on a free Vapi number, on a Twilio number you import, or over a SIP trunk from Telnyx, Plivo, DIDWW or another carrier. Every price below was read from Vapi’s and Twilio’s own pages on 18 November 2026, in US dollars.

Where Zian fits: an AI agents platform, not a developer toolkit

If what you actually want is calls answered and appointments booked, rather than a voice stack to own, Zian AI is the platform to look at first, and LeadsNow.ai is the done-for-you route. Zian AI is a standalone AI sales and reception agents platform, not a phone system (carrier / CPaaS / contact centre) and not an add-on to one.

Zian answers inbound calls 24/7 as an AI receptionist that qualifies, books, transfers and takes messages, runs outbound calls, and follows up by SMS, email, WhatsApp, Instagram DMs and Facebook Messenger, with answering-machine and voicemail handling on calls. SmartReach AI™ orchestrates the message, channel and timing of follow-up, and PrecisionPitch AI™ runs continuous split-testing of scripts and approaches. It supports 30+ languages and voice cloning, connects to HubSpot, Salesforce, HighLevel and Zapier, and offers private model deployment on customer infrastructure. Zian has set 64,521+ qualified sales appointments. LeadsNow.ai runs on Zian as a managed, pay-per-result service that answers leads in under 60 seconds, 24/7, and has set 50,769+ appointments. Zian is in a partnership-application beta with no published prices; each of its three listed plans, including Intel + Widgets (Free Forever), starts with a partnership application.

Apply For Partnership to see whether Zian’s inbound and outbound agents fit your call volume.

Vapi vs Twilio: what each one actually is on a phone call

Twilio is a communications platform (CPaaS). Its Programmable Voice product owns the phone number, the carrier connection and the call itself, and it bills per minute for making and receiving calls. On top of the call it sells two ways to reach an AI agent. Twilio’s ConversationRelay product page says it “connects AI agents to Twilio’s real-time voice AI infrastructure” and that you “get natural human-sounding TTS and real-time STT out of the box. Or, you can bring your own text-to-speech for a fully custom voice.” Its Media Streams overview gives “a real-time voice conversation with an AI assistant” as the example use for a bidirectional stream, where Twilio hands your server the audio and plays whatever audio you send back.

Vapi is a voice-agent orchestration platform. Its pricing FAQ says the per-minute cost has “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.” Vapi does not replace a carrier for outbound calling. Its free-number guide is plain: “Free Vapi numbers are inbound only. Import a number from a supported telephony provider to make outbound calls.” Importing a Twilio number takes your Twilio Account SID and Auth Token, after which Vapi’s guide says you can “assign it to an assistant for inbound calls or use it to place outbound calls.”

Question Twilio ConversationRelay Twilio Media Streams Vapi
What it is A TwiML noun that puts a live call on a WebSocket as text A TwiML noun that puts a live call on a WebSocket as raw audio A hosted agent platform: transcriber, model and voice configured in a dashboard or API
Who runs speech-to-text Twilio (Deepgram by default, or Google) You, with any vendor Vapi, at the provider you pick (calculator lists Deepgram, AssemblyAI, OpenAI)
Who runs text-to-speech Twilio (Google, Amazon or ElevenLabs), or your own TTS You, with any vendor Vapi, at the provider you pick (calculator lists ElevenLabs, Vapi voices, Cartesia)
Who runs the language model You (“bring-your-own-LLM”) You Vapi calls the model you configure (calculator lists OpenAI, Anthropic, Google)
What you host A WebSocket server that calls your LLM and tools A media server: audio, turn-taking, interruptions, STT, LLM, TTS A server URL for tool calls (Vapi’s FAQ: needed “for more advanced features like function calling”)
Phone numbers Twilio numbers ($1.15 a month local) Twilio numbers ($1.15 a month local) 1 free US number (inbound only) on Usage only; import Twilio, Telnyx or DIDWW numbers, or bring a SIP trunk
Published price on top of the call minute $0.07 a minute $0.0044 a minute $0.05 a minute hosting, plus STT, LLM and TTS at provider prices
Self-hosting Speech runs in Twilio’s cloud; your WebSocket server runs where you choose Your media server runs where you choose Vapi’s FAQ: Vapi “does not support on-premise deployments”

The one-sentence version: Twilio sells the call and the plumbing to an agent; Vapi sells the agent and, for outbound, still needs a carrier such as Twilio underneath.

What each published price includes, and what you still pay separately

Per-minute prices are only comparable once you know what each one covers. The table below prints each vendor’s own figure exactly as published, as at 18 November 2026, with what it includes and what sits outside it. Twilio’s US voice pricing page is marked “Pricing current as of October 2026”.

Published price Exactly as printed What it includes What you still pay separately
Twilio Programmable Voice, US local $0.0140 / min to make calls; $0.0085 / min to receive calls; local number $1.15 / mo The call leg on Twilio’s network Any AI product on top (ConversationRelay or Media Streams), recording ($0.0025 / min), answering machine detection ($0.0075 / call)
Twilio ConversationRelay $0.07 / minute (“starting at $0.07 per minute” in Twilio’s FAQ) The real-time voice connection to your agent; Twilio’s product FAQ says STT and TTS come “out of the box” “Voice costs are calculated separately” (the call minute above), your LLM, your WebSocket server
Twilio Media Streams $0.0044 / min Raw call audio to and from your WebSocket The call minute, your STT, LLM and TTS vendors, your media server and the code that runs turn-taking
Vapi hosting $0.05/min (“Pre-paid, $0.05/min Vapi hosting”) Orchestration of the call; models are “pass-through at cost” STT, LLM and TTS at provider prices; carrier minutes (“Charged by provider, not Vapi”); extra concurrency ($10 / line / month); success package if you want an SLA
Vapi transport Vapi Telephony / SIP: Free; Vapi WebSockets: Free; Daily WebRTC: Free; Twilio Inbound $0.008 / min; Twilio Outbound $0.014 / min; Vonage $0.00814 / min; Telnyx $0.0055 / min Free on a Vapi number or your own SIP trunk on the Vapi side Carrier minutes, billed by the carrier to your own account
Vapi Usage only $0 to start; $5 free credits; 1 included Vapi phone number; 4 concurrent calls Platform access, documentation, community support No response or uptime SLA; Core is $29 / month for 10 concurrent calls and a 2 business day response SLA

Two cells need a caution. First, Vapi’s transport table lists Twilio inbound at $0.008 a minute, while Twilio’s own page lists local inbound at $0.0085. Twilio bills those minutes, so the bills below use Twilio’s figure. Second, Twilio’s pricing pages show ConversationRelay as one $0.07 line and its product FAQ says speech recognition and voices come out of the box, but as at 18 November 2026 neither the voice pricing page nor the Conversations pricing page says in words that speech-to-text and text-to-speech are inside that $0.07. The same voice pricing page also says “Voice AI features have additional charges, such as running virtual agents with Google Dialogflow CX, virtual agent custom prompt voices, and text-to-speech” and lists premium text-to-speech at $0.0008–$0.0130 per 100 characters, without saying whether that applies to ConversationRelay. The bills below assume speech is inside the $0.07, as the live ConversationRelay vs Media Streams cost comparison also does. Confirm it on your first invoice.

What does a 10,000-minute month cost on Vapi vs Twilio?

This is the worked bill. Every line is a published rate multiplied by a quantity you control, so you can swap in your own numbers.

Assumptions, stated in full. 10,000 billed minutes a month of inbound calls to one US local number, the AI receptionist case, because that is the only case all four routes can serve (a free Vapi number cannot dial out). Peak of 4 simultaneous calls or fewer, so Vapi’s Usage only package needs no extra lines. No success package, no recording, no HIPAA add-on. The same model vendors on every route that runs its own models: Deepgram for speech-to-text, an OpenAI model and ElevenLabs voices. For those three lines we use Vapi’s published calculator bands on every route (Deepgram $0.00953–$0.00994, OpenAI $0.00774–$0.04523, ElevenLabs $0.0146–$0.02383 a minute), because Vapi says it bills them at the provider’s listed price with no markup. That keeps the model lines identical, so the difference between the columns is the platform and carrier, not our choice of model. The low column pairs the low end of every band, the high column the high end.

Line item, 10,000 inbound minutes A. Twilio ConversationRelay B. Twilio Media Streams + your models C. Vapi on an imported Twilio number D. Vapi on its free Vapi number
Platform fee 10,000 × $0.07 = $700.00 10,000 × $0.0044 = $44.00 10,000 × $0.05 = $500.00 10,000 × $0.05 = $500.00
Carrier minutes (Twilio local inbound $0.0085) $85.00 $85.00 $85.00 (billed by Twilio) $0.00 (Vapi Telephony listed as free)
Phone number $1.15 $1.15 $1.15 (Twilio) $0.00 (1 included)
Speech-to-text (Deepgram band) Inside the $0.07 (assumed) $95.30–$99.40 $95.30–$99.40 $95.30–$99.40
Language model (OpenAI band) $77.40–$452.30 $77.40–$452.30 $77.40–$452.30 $77.40–$452.30
Text-to-speech (ElevenLabs band) Inside the $0.07 (assumed) $146.00–$238.30 $146.00–$238.30 $146.00–$238.30
Monthly total, published prices $863.55–$1,238.45 $448.85–$920.15 $904.85–$1,376.15 $818.70–$1,290.00
Effective cost per minute $0.0864–$0.1238 $0.0449–$0.0920 $0.0905–$0.1376 $0.0819–$0.1290
Not priced: what you run yourself WebSocket server, LLM calls, tools Media server, turn-taking and interruption code, STT/LLM/TTS integration Tool-call server Tool-call server

How we checked it. Once line by line, as above, and once by summing per-minute rates first. Route A low: ($0.0085 + $0.07 + $0.00774) × 10,000 + $1.15 = $863.55. Route B low: ($0.0085 + $0.0044 + $0.00953 + $0.00774 + $0.0146) × 10,000 + $1.15 = $448.85. Route C high: ($0.05 + $0.0085 + $0.00994 + $0.04523 + $0.02383) × 10,000 + $1.15 = $1,376.15. Route D high: ($0.05 + $0.00994 + $0.04523 + $0.02383) × 10,000 = $1,290.00. Both methods agree to the cent in all eight cells. As a third check, Vapi’s own calculator shows “$82 – $129 /mo” for 1,000 minutes with Deepgram, OpenAI, ElevenLabs and Vapi telephony, which is route D at one tenth of the volume.

What the bill says. On published prices alone, building on Twilio Media Streams is the cheapest route at $448.85–$920.15, because Twilio’s audio pipe costs $0.0044 a minute against Vapi’s $0.05 hosting fee. ConversationRelay and Vapi on its own number land close together: route D runs from $44.85 cheaper to $51.55 dearer than route A, depending on which end of the model bands you use. Importing a Twilio number into Vapi (route C) is the dearest column for inbound, because you pay Twilio’s inbound minute and number on top of Vapi’s hosting.

If the same 10,000 minutes are outbound, Twilio’s $0.0140 outbound rate replaces the $0.0085 inbound rate, adding $55.00 to routes A, B and C: $918.55–$1,293.45, $503.85–$975.15 and $959.85–$1,431.15. Route D drops out, because a free Vapi number cannot place calls. Route C’s outbound figure matches the Vapi column in our Vapi vs Retell AI pricing breakdown, which uses the same inputs.

Two billing rules move these numbers in practice. Twilio’s support article on call duration 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, so a month of short calls bills more Twilio minutes than it talks. ConversationRelay is not on that list of five. We could not find a billing increment for Vapi’s hosting fee on its pricing page, its pricing and billing documentation or its changelog, as at 18 November 2026. And Vapi’s cost guide warns that its model figures are estimates: it lists “Long calls” as a case where “Actual cost is often higher”, because “Growing conversation history is sent again with each request.”

The Build-Budget Rule: what owning the pipeline has to cost less than

The Build-Budget Rule: multiply your monthly minutes by $0.0456. That is the most you can spend each month running your own media pipeline (servers, monitoring and the engineer who tunes turn-taking) before building on Twilio Media Streams costs more than running Vapi on the same Twilio number.

The $0.0456 is the gap between Vapi’s $0.05 hosting fee and Twilio’s $0.0044 Media Streams fee. On published prices every other line cancels: the Twilio minute, the number and the three model lines are the same on routes B and C. Vapi’s Twilio import guide does not say whether a Vapi-handled call adds any Twilio feature charge beyond the minute; if your first Twilio invoice shows one, add it to the $0.0456, because it makes route C dearer. So the rule holds for inbound and outbound alike, and for any model choice. At 10,000 minutes the build budget is $456.00 a month; at 50,000 minutes it is $2,280.00; at 100,000 minutes it is $4,560.00.

Two variants cover the other routes. Against a free Vapi number on inbound calls, the gap is smaller, because Vapi carries the call for free: $0.0371 a minute ($0.05 − $0.0085 − $0.0044), less the $1.15 Twilio number, or $369.85 at 10,000 minutes. Against ConversationRelay there is almost no price gap to fund anything: routes A and D sit within about $52 of each other at 10,000 inbound minutes. If the choice is between ConversationRelay and Vapi, decide on what you want to own, not on price.

Monthly minutes Build budget vs Vapi on a Twilio number (× $0.0456) Build budget vs a free Vapi number, inbound (× $0.0371 − $1.15) What that budget has to cover on Media Streams
2,000 $91.20 $73.05 Hosting alone; no engineering time left
10,000 $456.00 $369.85 Hosting plus a few hours a month of maintenance
50,000 $2,280.00 $1,853.85 Hosting plus part of an engineer
100,000 $4,560.00 $3,708.85 Hosting plus a meaningful share of an engineer

The third column is our reading, not a vendor figure; replace it with your own hosting quote and loaded engineering rate. If you also need concurrency above Vapi’s 4 included lines, add $10 a line a month to the Vapi side, which raises the build budget by the same amount. Our Vapi vs LiveKit Agents comparison shows how peak simultaneous calls, not minutes, decide which Vapi package you need.

Which route fits which team

The table puts the platform answer first, then the four build routes. It is a fit guide, not a scored ranking: the Vapi and Twilio rows come from their own pages, and the Zian and LeadsNow.ai rows from what those companies publish. zian.ai publishes this page and appears in it.

If you are… Route Why Wrong for you if…
A business that wants calls answered and appointments booked, not a voice stack Zian AI (platform); LeadsNow.ai for done-for-you AI receptionist and outbound agents with SMS, email, WhatsApp, Instagram DM and Messenger follow-up in one standalone platform; LeadsNow.ai runs it for you, pay-per-result You need to write your own agent code or need a published price today
A developer team that wants a dashboard, tools and outbound campaigns without owning audio C or D: Vapi Transcriber, model and voice are settings; inbound can start on a free Vapi number You must run on premises (Vapi’s FAQ says it does not) or need outbound on a free number
A Twilio shop whose agent thinks in text A: Twilio ConversationRelay Twilio runs STT, TTS and barge-in; your server only trades text with your LLM You need a speech-to-text or voice vendor Twilio’s TwiML reference does not list (it names Deepgram and Google; Google, Amazon and ElevenLabs), although its product page also mentions bringing your own TTS
A team with engineers to own audio, at volume B: Twilio Media Streams Lowest published per-minute total, any model vendor Your monthly build cost exceeds minutes × $0.0456
A team that wants Twilio as the carrier but Vapi as the agent C: Vapi on an imported Twilio number Keeps the number on your own Twilio account (a free Vapi number cannot be ported out); adds outbound The inbound-only job could run on a free Vapi number for less

Vapi also documents a Twilio SIP trunk integration and guides for Telnyx, Plivo, DIDWW and other trunks, so “Vapi on Twilio” does not have to mean importing a number. If you want neither a Twilio media product nor a platform in the call path, our guide to running an AI voice agent on direct SIP with no Twilio in the path covers that route.

Rather not build on Vapi or Twilio at all? Zian runs AI receptionist, outbound calling and messaging agents as one platform. Apply For Partnership

The sentence on each vendor’s pages that cuts against this page

On Vapi’s side it is the billing guide’s “Every minute your assistants run is billed”, read with the $0.05 hosting fee. At the published rate, a team that grows to 100,000 minutes pays $5,000 a month in hosting before any model, which is the whole case for route B. Vapi’s FAQ does say “volume pricing is available as you scale”, without a published figure, and Pro is priced at 10% of Vapi hosting with a $999 a month minimum on top of usage. Vapi’s Usage only package also has no response or uptime service-level agreement; a response SLA starts with Core and a 99% uptime SLA with Pro.

On Twilio’s side it is the ConversationRelay pricing line “Voice costs are calculated separately”, together with the FAQ’s “starting at $0.07 per minute”. ConversationRelay is not the whole bill: the call minute, the number and your language model all sit outside it, and Twilio rounds the call minute up to the next whole minute. Twilio’s own voice FAQ concedes the price point before arguing for it: “You can find vendors who charge less upfront, but when you build with Twilio, you get a premium offering that works reliably and delivers lasting ROI over time.” Route B is cheap on paper because the expensive part, the media server and the people who keep turn-taking working, is not on any price list.

What this bill leaves out

The worked bill prices the published per-minute lines and nothing else. It leaves out your servers on every route, engineering time, taxes, recording and storage, answering machine detection, Vapi’s HIPAA add-on ($2,000 a month), Vapi concurrency above 4 lines, Twilio volume discounts, and any committed-use pricing either vendor negotiates. Vapi’s model bands are Vapi’s estimates, built on 5 model requests per turn, 150 output tokens a minute and a 50% prompt-cache hit rate; Vapi’s transcription estimate also counts two audio channels (caller and assistant), and a pipeline you build on Media Streams might transcribe only the caller, which would lower route B further. We could not find, on Vapi’s import or telephony guides, whether Vapi’s Twilio import adds any Twilio feature charges beyond the call minute, as at 18 November 2026; check the first Twilio invoice after you import a number.

Frequently asked questions

Is Vapi cheaper than Twilio for an AI voice agent?

Not on published prices. At 10,000 inbound US minutes we itemise $448.85–$920.15 on Twilio Media Streams with your own models, against $818.70–$1,290.00 on Vapi with its free number. Twilio ConversationRelay comes to $863.55–$1,238.45. The Media Streams figure leaves out the servers and engineers you would need.

Does Vapi use Twilio?

It can. Vapi’s pricing page lists Twilio inbound at $0.008 and outbound at $0.014 a minute as transport “Charged by provider, not Vapi”, and you can import a Twilio number with your Account SID and Auth Token. Vapi also offers free Vapi numbers and SIP trunks from Telnyx, Plivo and other carriers.

Can a free Vapi number make outbound calls?

No. Vapi’s free phone number guide says “Free Vapi numbers are inbound only. Import a number from a supported telephony provider to make outbound calls.” Free numbers use US area codes, cannot launch Outbound Campaigns and cannot be ported out.

How much does Twilio ConversationRelay cost per minute?

Twilio lists Conversation Relay at $0.07 a minute (“starting at”), with the call minute billed separately: $0.0085 to receive or $0.0140 to make a US local call. Its product FAQ says speech-to-text and text-to-speech come out of the box; your LLM is extra.

What is the Build-Budget Rule for Vapi vs Twilio?

Multiply your monthly minutes by $0.0456, the gap between Vapi’s $0.05 hosting fee and Twilio’s $0.0044 Media Streams fee. If owning your own media pipeline costs less than that each month, Twilio Media Streams is cheaper than Vapi on the same Twilio number. At 10,000 minutes the budget is $456.

Does Twilio round call minutes up?

Yes. Twilio’s call duration billing article says 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. We could not find a billing increment for Vapi’s hosting fee.

Where every figure on this page comes from

Figure Who published it Link Date read
$0.05/min hosting; models “pass-through at cost”; four components at provider price, no markup Vapi vapi.ai/pricing 18 November 2026
Transport: Vapi Telephony / SIP free; Twilio $0.008 in, $0.014 out; Vonage $0.00814; Telnyx $0.0055 Vapi vapi.ai/pricing 18 November 2026
Calculator bands: Deepgram $0.00953–$0.00994, OpenAI $0.00774–$0.04523, ElevenLabs $0.0146–$0.02383; “$82 – $129 /mo” at 1,000 minutes Vapi vapi.ai/pricing 18 November 2026
Usage only: $5 credits, 1 number, 4 concurrent calls; Core $29/month, 10 calls; Pro 10% of hosting, $999/month minimum; $10/line/month; HIPAA $2,000/month Vapi vapi.ai/pricing 18 November 2026
Free Vapi numbers inbound only, US area codes, no Campaigns, no port-out Vapi docs.vapi.ai/free-telephony 18 November 2026
Import a Twilio number with Account SID and Auth Token; inbound and outbound Vapi docs.vapi.ai/phone-numbers/import-twilio 18 November 2026
“Every minute your assistants run is billed”; no SLA on Usage only Vapi docs.vapi.ai/billing/pricing-and-success-packages 18 November 2026
5 requests per turn; 150 output tokens a minute; 50% cache; two audio channels; long calls cost more Vapi docs.vapi.ai/assistants/model-intelligence/understanding-cost 18 November 2026
No on-premise deployments; server URL for function calling Vapi docs.vapi.ai/faq 18 November 2026
US local $0.0140/min out, $0.0085/min in, $1.15/month number; Media Streams $0.0044; Conversation Relay $0.07; recording $0.0025/min; AMD $0.0075/call; premium text-to-speech $0.0008–$0.0130 per 100 characters; “Voice AI features have additional charges” Twilio twilio.com/en-us/voice/pricing/us 18 November 2026
Conversation Relay $0.07/minute, “starting at”; “Voice costs are calculated separately” Twilio twilio.com/en-us/products/conversational-ai/pricing 18 November 2026
STT and TTS “out of the box”; bring your own LLM Twilio twilio.com/en-us/products/conversational-ai/conversationrelay 18 November 2026
ConversationRelay providers: Deepgram or Google STT; Google, Amazon or ElevenLabs TTS Twilio twilio.com/docs/voice/twiml/connect/conversationrelay 18 November 2026
Partial minutes rounded up; five products listed Twilio help.twilio.com/articles/223132307 18 November 2026
All route totals, the $0.0456 and $0.0371 gaps, and the build-budget table Zian (our arithmetic from the rows above) This page 18 November 2026

Vapi and Twilio change prices often; every figure here is as at 18 November 2026. If you would rather run answered calls and booked appointments than price a voice stack, Apply For Partnership.

Related Blogs

Related from Zian AI