Vapi AI Pricing (2026): Plans, Real Costs and What You'll Actually Pay

Prices checked September 19, 2026 on vapi.ai/pricing. Hosting and provider usage are billed on every package; a package adds support and capacity on top.

For what Vapi is and how its pipeline works, see our Vapi AI complete guide. This page is only about the bill.

How a Vapi minute is built

Vapi's pricing FAQ names four components on every minute: speech-to-text, the language model, text-to-speech and transport, "each billed at the provider's listed price with no markup," with the $0.05 hosting fee added on top. Using the defaults on Vapi's calculator:

Transport is only free on Vapi's own numbers and connections. If you bring a carrier, the carrier bills you: the pricing page lists Twilio at $0.008 a minute inbound and $0.014 outbound, Vonage at $0.00814 and Telnyx at $0.0055.

The model line is the one you control most, and it depends on more than which model you pick. Vapi's cost methodology page says the estimate is built from your actual system prompt and tool definitions, assumes 5 model requests per conversational turn and about 150 output tokens a minute, and blends in cached-token rates at an assumed 50% cache hit rate. Two agents on the same model can therefore cost different amounts a minute: every tool definition you add is sent on every request.

You can also skip Vapi's pass-through entirely for a provider. With provider keys, "you won't be charged when using that provider through Vapi. Instead, you'll be charged directly by the provider." That's how a team with an enterprise OpenAI or ElevenLabs contract gets its negotiated rate while still paying Vapi's $0.05 hosting.

What Vapi costs at three call volumes

We ran Vapi's calculator at three volumes with its default providers, then added the cheapest package that covers each volume's concurrency. The concurrency figures are our assumptions for an inbound support line: peaks under 4 simultaneous calls at 1,000 minutes, 8 at 10,000 and 25 at 50,000.

Per minute, the all-in range barely moves with volume: $0.082 to $0.129 at the small size and $0.085 to $0.133 at the large one, once the concurrency lines are in. Vapi's FAQ says "volume pricing is available as you scale" but doesn't publish it, so these are list prices.

When Pro is worth $999

Pro is priced at 10% of your Vapi hosting fees with a $999 monthly minimum, and the FAQ confirms "model costs are not included" in that percentage. At $0.05 a minute, 10% of hosting is $0.005 a minute, so Pro only rises above its $999 floor past 199,800 minutes a month. Below that, Pro is a flat $999.

At our large volume, Pro costs $820 more than Core plus 15 concurrency lines, and it buys things a line add-on can't: a dedicated Slack channel, a one-business-day response SLA, a 99% uptime SLA, 180 days of call data, basic roles and 10 organizations. Pro is a support contract. If your agents take calls that can't go unanswered, that's what the $820 buys. If they don't, Core and add-on lines get you the same capacity.

Add-ons and the HIPAA line

Vapi sells three add-ons on any package, including Usage only:

• HIPAA compliance: $2,000 a month per organization, with a standard DPA and BAA.
• Additional concurrency: $10 per concurrent call line a month.
• Additional organizations: $20 per organization a month.

HIPAA changes the shape of a small bill completely. At 1,000 minutes a month, a HIPAA organization costs $2,082 to $2,129, and 94% to 96% of that is the add-on. The fee is per organization, and Vapi suggests using organizations to separate dev, staging and production, so a team that wants a HIPAA-covered staging environment pays for it twice. On Trustpilot, a reviewer wrote in April 2026 that Vapi charged "$1000 for a HIPPA compliant voice receptionist"; the pricing page now lists $2,000.

How Vapi bills: credits, freezes and packages in the dashboard

Usage runs on prepaid credits. Per Vapi's billing docs, the hosting fee and provider costs draw down your balance as calls run, and "if your credit balance reaches $0 and auto-reload is off, Vapi freezes the subscription." For a production line, auto-reload isn't optional.

Packages are bought in the dashboard's Packages tab, and the fee is prorated to the start of the next billing month. The purchase dialog in Vapi's docs shows Core raising concurrent lines from 4 to 10 and numbers from 1 to 5, with part of the prorated fee paid from existing credit and the rest charged to the card.

Two details trip up new accounts. Free Vapi numbers are US-only and inbound-only, so outbound calling and campaigns need an imported Twilio, Telnyx or Vonage number, and that carrier then bills per minute. And chat has its own gate: the Chat API requires a card on file, and a pay-as-you-go organization under 30 days old must have completed a credit purchase before chat works. The pricing page prices voice minutes; it lists no separate per-message rate for chat.

Success Packages apply to new customers. Vapi's packages page says existing accounts keep their current hosting fee and features for now, which is why some third-party pages still quote older terms such as 10 concurrent calls or $10 of starting credit.

Enterprise pricing: AWS Marketplace and Vendr

Vapi sells through AWS Marketplace, where the listing "Voice AI for developers" has one usage dimension: $0.05 per minute engaged on a call, matching the hosting fee on Vapi's site. Buying there puts the hosting fee on your AWS bill; provider costs still apply.

For negotiated contracts, Vendr, a software-buying marketplace that doesn't compete with Vapi, reports a median of $105,000 a year from its buyers' data, with a range of $60,000 to $105,000. Vendr describes the enterprise tier as a fixed annual platform license plus a prepaid usage bank. That figure covers contracts, not the self-serve price, and Vendr's description of the entry tier is out of date against Vapi's current page.

What drives a Vapi bill up

• The model and the prompt. On the calculator's default OpenAI range, the model is $0.0077 to $0.0452 a minute, a sixfold spread. Long system prompts and many tools raise it further, because Vapi's estimate (and your provider's bill) counts input tokens on every request.
• The voice. ElevenLabs runs $0.0146 to $0.0238 a minute on the calculator, billed per character spoken, so an agent that talks more costs more.
• Outbound and international calls. Free numbers can't dial out. Twilio outbound is $0.014 a minute, and an international number means importing one from a carrier.
• Concurrency. Usage only includes 4 lines. Every line past your package is $10 a month.
• Compliance. HIPAA is $2,000 a month per organization, at any volume.
• Environments. Each organization past your package's 2 (or 10 on Pro) is $20 a month.

What Vapi's pricing rewards

Vapi's markup sits entirely on the hosting fee and the packages. Provider costs pass through at list price, so Vapi earns the same $0.05 a minute whether your agent runs a cheap model or an expensive one. That's an unusually clean incentive: Vapi has no reason to steer you to a pricier model, and the calculator shows the provider ranges openly.

What Vapi does earn more from is minutes and support. Usage only comes with Discord and no SLA, and the step up to a response commitment is a package. Pro's price is a percentage of your hosting fees, so support costs scale with usage past 199,800 minutes. Given that the most common complaint in Vapi's public reviews is support, the packages are where Vapi prices the thing its users say they're missing.

How Vapi's price compares

At the large volume, 50,000 minutes a month, with each platform's cheapest option for 25 simultaneous calls:

Rival prices are from each vendor's live page on September 19, 2026.

Retell prices each component itself rather than passing providers through. Its calculator gives $5,450 at 50,000 minutes on GPT 5.4 mini; a knowledge base, phone numbers and 5 lines above its free 20 bring it to about $5,750. Retell includes 20 concurrent calls free, which is what makes it simpler at mid volumes. Our Retell AI guide covers the product.

Bland quotes one rate with the model included: $0.12 a minute on Build, plus $299 a month, which lifts the daily cap to 2,000 calls. It's the most predictable number here and the most expensive at this volume. See our Bland AI guide.

ElevenLabs Agents bills $0.08 for every minute past a plan's allowance, so Pro, Scale and Business all land within a cent of $4,000 at 50,000 minutes. Scale is the cheapest one with 30 concurrent calls. The model is added per minute and telephony is billed separately. Our ElevenLabs Agents guide goes further.

On list price, Vapi's low end and ElevenLabs are the cheapest options at this size, and Vapi's high end is close to Bland's. Where Vapi wins is choice: it passes provider costs through at list price, and with provider keys you can pay your own negotiated rates for the transcriber, model and voice. For contact centers that would rather buy a finished product than assemble one, Parloa sells on contract, and Voiceflow suits teams that design flows visually.

What users say about Vapi's pricing

Vapi has too few reviews on G2 to lean on: 3 out of 5 from 3 reviews when we checked in Chrome on September 19, 2026. Trustpilot's Vapi page has more: 2.4 from 15 reviews, 8 of them in the last 12 months. The profile is unclaimed and Vapi hasn't invited reviews there, so the sample skews to people with a complaint.

The pricing complaints are specific:

• Isadora (Germany, December 2025) called the headline misleading: "its just for platform if you add LLM's it will cost High same goes with Twilio," putting her average at "0.17 to 0.50 depends on your setting."
• Sam Hardwick (UAE, April 2026): "They charge $1000 for a HIPPA compliant voice receptionist. Others offer that for a fraction of the price."
• A reviewer in the UK (October 2024) on numbers: without UK numbers to buy, "you will pay twice to make one call."

The first complaint is the one this page exists to answer. Vapi's per-minute price is a platform fee, and the calculator shows the rest, but a builder who budgets from $0.05 will be off by a factor of two or more. Reviewers' $0.17 to $0.50 is higher than the calculator's $0.08 to $0.13, which fits premium voices, larger models, Twilio telephony and long prompts, but we can't check any single reviewer's setup.

Who Vapi's pricing suits

Vapi's pricing fits developers who want to choose every provider, already hold negotiated model or voice contracts, or need to run the cheapest possible model at volume. At 1,000 minutes it's one of the cheapest options on this page, and at 50,000 it stays competitive if you keep the model lean. It's a harder sell for regulated teams at low volume, where the $2,000 HIPAA add-on dominates the bill, and for teams that need guaranteed support but don't want to spend $999 a month on Pro.

If your support volume is mostly email and chat tickets, a voice pipeline is the wrong purchase. Macha is an AI agent layer on the help desk you already run (Zendesk, Freshdesk, Gorgias, Front, HubSpot or Intercom), billed per ticket: one thread with one person is one charge however many replies it takes, from $299 a month for 750 tickets, with setup and monitoring by the Macha team included. The tiers are on our pricing page, and our guide to AI agents for customer service covers where voice and ticket agents each fit.

How we researched this

We read Vapi's pricing page with a scripted browser on September 19, 2026, including the add-ons, transport fees and the expanded FAQ, and ran Vapi's own calculator at 1,000, 5,000, 10,000 and 50,000 minutes for the figures and screenshot above. Billing rules come from Vapi's docs on packages, add-ons, credits, cost estimation, provider keys, concurrency, free numbers and chat; the two dashboard images are from those docs. We read the AWS Marketplace listing and G2 in Chrome, Vendr and Trustpilot with a scripted browser, and priced each rival from its own page the same day. We did not open a Vapi account or build an agent for this post.

Frequently asked questions

How much does Vapi cost per minute? Vapi's hosting fee is $0.05 a minute. With its calculator's default providers (Deepgram, OpenAI, ElevenLabs) and Vapi telephony, a minute costs $0.0818 to $0.1289 all in, and the model accounts for most of the range.

How much does Vapi cost per month? There's no required monthly fee on Usage only. At list prices, 1,000 minutes costs $82 to $129 a month, 10,000 minutes $848 to $1,319 with the $29 Core package, and 50,000 minutes $4,273 to $6,629 with Core and 15 extra concurrency lines.

Is Vapi free or paid? Paid. New accounts get $5 of credit and one free US inbound number, which covers roughly 40 to 60 minutes at the calculator's rates. After that every minute is billed from prepaid credit.

Is Vapi AI expensive? On list price it's in the middle of its market. At 50,000 minutes, Vapi's range ($4,273 to $6,629) sits beside ElevenLabs Agents (about $4,060 plus telephony), Retell (about $5,750) and Bland ($6,299, with telephony possibly extra). It feels expensive when you budget from the $0.05 headline, which covers hosting only.

Why is Vapi so expensive for HIPAA? HIPAA-eligible handling is a flat $2,000 a month per organization, with a DPA and BAA. It doesn't scale with minutes, so it dominates a small deployment: at 1,000 minutes it's about 95% of the bill.

What does Vapi Pro cost? Pro is 10% of your Vapi hosting fees with a $999 monthly minimum, model costs excluded. Because 10% of hosting is $0.005 a minute, Pro stays at $999 until you pass 199,800 minutes a month.

Is there a Vapi pricing calculator? Yes, on vapi.ai/pricing. It covers hosting, transcriber, model, voice, transport and package, but not concurrency or organization add-ons, HIPAA, or carrier charges on imported numbers.

How does Vapi's pricing compare to Retell's? Vapi adds $0.05 a minute to providers at cost; Retell prices each component itself, starting from $0.055 a minute of voice infrastructure. At 50,000 minutes our Retell configuration comes to about $5,750 and Vapi to $4,273 to $6,629. Retell includes 20 concurrent calls free where Vapi's Usage only includes 4.

What does Vapi cost for enterprise? Premier is priced by contract. Vendr reports a median Vapi contract of $105,000 a year, with a range of $60,000 to $105,000, and describes it as a platform license plus a prepaid usage bank.

If your team answers tickets rather than phone calls, you can start a Macha trial with $50 of free usage and no card, or see the tiers on our pricing page.

Sources: Vapi pricing · Vapi: Pricing and Success Packages · Vapi: manage packages and add-ons · Vapi: manage billing and credits · Vapi: how component costs work · Vapi: provider keys · Vapi: free phone numbers · Vapi: call concurrency · Vapi: chat quickstart · Vapi FAQ · AWS Marketplace: Vapi · Vendr: Vapi · Trustpilot: Vapi (trustpilot.com/review/vapi.ai) · G2: Vapi · Retell pricing · Bland pricing · ElevenLabs Agents pricing