Vapi Pricing 2026: Per-Minute Costs, Plans & Alternatives

Vapi charges per-minute usage with no published base subscription fee, starting at $0.05 per minute for basic calls on the Developer plan, though costs scale with model and voice selection; this article breaks down the pricing structure, what's included, and how to estimate your bill.

Major takeaways

How much does Vapi cost? Vapi's Developer plan starts at $0.05 per minute for basic voice calls, with costs rising to $0.15–$0.30+ per minute when using premium LLMs or voices; enterprise pricing is custom and requires a sales conversation.

What's included in the per-minute rate? The per-minute rate reportedly covers platform access, telephony, speech-to-text, text-to-speech, and LLM inference, though specific model tiers and voice providers carry different rates within that structure.

What else should you budget for? Phone number fees (typically $1–$2 per number per month), premium voice or LLM upgrades, engineering time to build and maintain call flows, and integration work with CRMs or calendars add to the headline per-minute cost.

Quick answer box

What Vapi is

Vapi is a developer-focused voice AI platform for building conversational phone agents. Teams use it to handle inbound calls (customer support, appointment booking, lead qualification) and outbound dialing (reminders, surveys, follow-ups). The platform provides APIs for speech-to-text, LLM orchestration, text-to-speech, and telephony in a single stack. Developers write call flows in code or use Vapi's dashboard to configure agents without managing separate STT, TTS, or telephony vendors.

Vapi positions itself for engineering teams that want full control over call logic and model selection. It supports multiple LLM providers (OpenAI, Anthropic, custom endpoints) and voice engines (ElevenLabs, Deepgram, Play.ht). The platform handles both inbound and outbound calls, with features for call transfer, voicemail detection, and real-time interruption handling.

Customers range from startups automating receptionist tasks to contact centres routing high-volume support calls.

How Vapi pricing works

The pricing model

Vapi uses a per-minute usage model with no upfront platform fee on the Developer plan. You pay for the minutes your agents spend on calls, with the rate determined by the combination of LLM, voice provider, and telephony options you select. The Developer plan is pay-as-you-go; the Enterprise plan moves to custom pricing with volume discounts and annual contracts.

There is no published seat-based or subscription tier between these two.

New accounts receive $10 in free credits to test the platform. After that, you add funds to your account and Vapi deducts usage in real time. The per-minute rate is composite: it bundles the cost of telephony (inbound or outbound), speech-to-text transcription, LLM inference, and text-to-speech synthesis into a single line item.

The rate varies based on which models and voices you choose.

What the per-minute rate covers

The per-minute rate includes platform access, telephony (SIP trunking for inbound and outbound calls), speech-to-text transcription, LLM inference, and text-to-speech synthesis, as listed on Vapi's pricing page at time of writing. The baseline $0.05 per minute rate applies when using Vapi's default or lower-cost model and voice configurations. Upgrading to premium LLMs (GPT-4, Claude 3.5 Sonnet) or premium voices (ElevenLabs Turbo, Play.ht ultra-realistic) increases the per-minute cost to $0.15–$0.30 or higher.

Telephony costs are embedded in the per-minute rate. Vapi does not charge separately for inbound or outbound minutes beyond the composite rate.

The pricing page does not specify whether international calling incurs additional fees; buyers should confirm with Vapi's support whether calls to non-US numbers carry surcharges.

Speech-to-text and text-to-speech are included, but the choice of provider affects the rate. Deepgram STT is reportedly less expensive than Whisper; ElevenLabs voices cost more than standard TTS engines. The LLM component scales with token usage, so longer or more complex conversations drive higher per-minute costs even if the wall-clock duration is the same.

Limits and add-ons

The Developer plan has no published concurrency limit, though Vapi's documentation notes that high-concurrency use cases (100+ simultaneous calls) should contact sales for capacity planning. Phone numbers are billed separately at approximately $1–$2 per number per month, depending on the region and whether the number supports SMS.

Toll-free numbers typically cost more than local numbers.

Premium voice models and LLMs are add-ons in the sense that selecting them raises the per-minute rate. There is no separate line-item charge; the cost is embedded in the usage billing. Vapi does not publish a support tier structure for the Developer plan. Enterprise customers reportedly receive dedicated support and SLAs as part of their custom contracts.

The platform does not charge for failed calls (calls that do not connect) or for time spent in voicemail detection, according to third-party user discussions, though this is not explicitly confirmed on the pricing page. Buyers running high-volume outbound campaigns should verify how Vapi meters unsuccessful dial attempts.

Plan breakdown

| Plan | Price | Included | Best for | |---|---|---| | Developer | $0.05–$0.30+/min usage | Platform access, telephony, STT, LLM, TTS; $10 free credits | Startups, dev teams testing or running low-to-mid volume agents | | Enterprise | Custom quote | Volume discounts, dedicated support, SLAs, capacity planning | High-volume contact centres, sales teams with 10,000+ calls/month |

Developer plan: Pay-as-you-go with no minimum spend. The $0.05 per minute baseline applies to basic configurations (standard LLM, standard voice). Costs rise when you select premium models or voices. You add funds to your account and Vapi deducts usage in real time.

This plan suits teams running a few hundred to a few thousand calls per month and comfortable managing their own infrastructure. There is no published concurrency cap, but Vapi recommends contacting sales if you expect sustained peaks above 100 simultaneous calls.

Phone numbers are billed separately at $1–$2 per number per month.

Enterprise plan: Custom pricing based on committed volume, use case, and support requirements. Vapi's sales team negotiates per-minute rates lower than the Developer tier when you commit to a minimum monthly spend or annual contract. Enterprise customers reportedly receive dedicated account management, priority support, and SLA guarantees.

This plan is the starting point for contact centres, large sales teams, or any buyer running more than 10,000 calls per month. The pricing page does not publish example Enterprise rates; expect to provide call-volume forecasts and use-case details during the sales process.

Estimate your monthly cost

The formula for estimating your monthly Vapi bill is:

Monthly cost ≈ (calls per month × average minutes per call × all-in per-minute rate) + phone number fees + add-ons

Each input:

• Calls per month: Total number of calls your agents handle (inbound) or place (outbound). For outbound, multiply dials by your connect rate (the percentage of calls that reach a live person). A 30% connect rate on 1,000 dials means 300 connected calls.
• Average minutes per call: Measured from the moment the call connects to the moment it ends. Include hold time, transfers, and any time the agent is speaking or listening. Voicemail detection time is reportedly not metered, but confirm with Vapi.
• All-in per-minute rate: The composite rate for your chosen LLM, voice, and telephony. Use $0.05 for baseline, $0.15 for mid-tier (e.g., GPT-4 + standard voice), $0.30 for premium (Claude 3.5 + ElevenLabs Turbo). If you have a custom Enterprise rate, substitute that.
• Phone number fees: $1–$2 per number per month. Add more numbers if you need separate lines for different campaigns or regions.
• Add-ons: Integration costs, QA tools, compliance recording, or premium support (if applicable).

The table below shows three illustrative scenarios. Every input is an assumption; replace these with your own call data.

The small scenario assumes a single inbound receptionist agent handling 500 calls per month at 3 minutes each with a basic LLM and voice ($0.05/min). The mid scenario models a sales team qualifying 2,000 inbound leads per month at 5 minutes each with a mid-tier configuration ($0.15/min). The high scenario represents a contact centre running 10,000 calls per month at 8 minutes each with premium models ($0.30/min).

These are not benchmarks; your costs will differ based on call length, model choice, and connect rates. Outbound campaigns should factor in dial attempts that do not connect (if metered) and retries.

Hidden and adjacent costs

Telephony and phone number fees

Phone numbers are billed separately from the per-minute rate. Local numbers typically cost $1–$2 per month; toll-free numbers run $2–$5 per month, depending on the region. If you need multiple numbers for different campaigns, geographies, or A/B tests, multiply accordingly.

International numbers or calls to non-US destinations may carry surcharges; Vapi's pricing page does not specify international rates, so confirm with support before launching global campaigns.

LLM and voice model upgrades

Selecting a premium LLM (GPT-4, Claude 3.5 Sonnet) or a premium voice (ElevenLabs Turbo, Play.ht ultra-realistic) raises the per-minute rate from the $0.05 baseline to $0.15–$0.30 or higher. The cost scales with token usage, so longer conversations or agents that generate verbose responses drive higher bills even if the call duration is the same.

If your use case requires low latency or high-quality voice, budget for the upper end of the per-minute range. Teams running cost-sensitive campaigns should test whether a mid-tier LLM and standard voice deliver acceptable results before committing to premium tiers.

Engineering time to build and maintain flows

Vapi is a developer platform. Building a production-ready agent requires writing call-flow logic, integrating with your CRM or calendar, handling error states, and testing edge cases (background noise, accents, interruptions). Expect 40–80 hours of engineering time to launch a basic agent and 10–20 hours per month to maintain and iterate.

If your team lacks in-house voice AI experience, add onboarding time or budget for a consultant. Vapi provides API documentation and example code, but it does not offer a no-code builder for non-technical users.

Integration costs

Connecting Vapi to your CRM (Salesforce, HubSpot), calendar (Google Calendar, Calendly), helpdesk (Zendesk, Intercom), or data warehouse requires custom API work. Vapi supports webhooks and REST endpoints, but you will need to write the integration logic and handle authentication, error handling, and data mapping.

If you use a third-party integration platform (Zapier, Make), expect to pay for those subscriptions separately. Budget 20–40 hours of engineering time per integration, plus ongoing maintenance when APIs change.

QA and call review

Vapi provides call logs and transcripts, but it does not include a built-in QA dashboard for scoring agent performance or flagging problematic calls. Teams running high-volume agents typically export call data to a separate analytics tool or build custom dashboards.

If you need human review of a sample of calls (common in regulated industries or sales environments), budget for the time or a third-party QA service. Expect to review 5–10% of calls in the first month and 1–2% ongoing once the agent is stable.

Compliance and data retention

If your use case requires call recording for compliance (finance, healthcare, insurance), confirm that Vapi's recording and storage meet your regulatory requirements. Vapi stores call logs and transcripts, but the pricing page does not specify retention periods or whether long-term storage incurs additional fees.

Teams subject to GDPR, HIPAA, or TCPA should verify that Vapi's data handling aligns with their obligations. If you need to store recordings beyond Vapi's default retention, budget for external storage (S3, Google Cloud Storage) and the engineering time to export and archive call data.

Is Vapi worth the price?

Vapi delivers strong value for engineering teams that want full control over call logic and model selection without managing separate STT, TTS, and telephony vendors. The per-minute model scales predictably with usage, and the $0.05 baseline rate is competitive for basic configurations.

Teams running a few thousand calls per month with standard LLMs and voices will find the Developer plan straightforward and cost-effective.

The pricing becomes less attractive when you need premium models or high call volumes. A contact centre running 10,000 calls per month at 8 minutes each with premium configurations will pay $24,000+ per month, which is comparable to or higher than hiring a small team of human agents in some markets. At that scale, Enterprise pricing with volume discounts becomes necessary, but Vapi does not publish those rates, so buyers cannot model costs without a sales conversation.

Public review volume for Vapi is limited; the recurring themes below are drawn from the available G2 / Trustpilot / TrustRadius coverage. Users praise the platform's flexibility and API-first design, noting that it allows them to build agents tailored to specific workflows without vendor lock-in. The developer experience is reportedly strong, with clear documentation and responsive support.

The main trade-off is that Vapi requires engineering resources; non-technical teams will struggle to deploy agents without hiring developers or consultants.

Vapi suits startups and mid-market companies with in-house engineering capacity that want to build custom voice agents for inbound or outbound use cases. It is less suitable for teams that need a no-code builder, pre-built agent templates, or a platform that handles integrations out of the box.

Buyers running high-volume campaigns should compare Enterprise pricing against alternatives that publish volume-discount tiers.

Vapi alternatives to price against

Retell AI charges per-minute usage starting at $0.08 per minute for basic configurations, with costs rising for premium models and voices. The platform is developer-focused like Vapi, with strong support for custom LLM endpoints and low-latency voice. Retell is worth comparing if you need sub-500ms response times or plan to use proprietary models.

The pricing structure is similar to Vapi's; the main differentiator is Retell's emphasis on latency optimization.

Bland AI starts at $0.09 per minute on the standard plan and includes pre-built templates for outbound sales, appointment setting, and lead qualification. Bland is less flexible than Vapi but faster to deploy for teams that fit its use-case templates. It is a better starting point for sales teams that want to launch agents quickly without custom development.

Bland's pricing is slightly higher than Vapi's baseline but includes more out-of-the-box functionality.

11x's Julian is an AI phone agent for inbound sales, sold as part of the 11x platform alongside Alice (AI SDR for outbound). Pricing is custom and requires a demo. Julian is built for speed-to-lead (answering inbound calls in under 10 seconds) and qualification, with native integration to 11x's outbound engine.

It suits revenue teams that want a unified platform for inbound voice and outbound email/LinkedIn, rather than a standalone voice API. Compare Julian if you are already evaluating AI SDRs and want inbound voice in the same system.

Synthflow charges $29 per month plus $0.10 per minute and provides a no-code builder for simple inbound agents. It is designed for non-technical teams and includes pre-built integrations with common CRMs and calendars. Synthflow is worth comparing if you lack engineering resources or need to deploy an agent in days rather than weeks.

The per-minute rate is higher than Vapi's, but the no-code builder reduces the total cost of ownership for small teams.

How to negotiate

Commit to volume for a lower per-minute rate

Vapi's Enterprise plan offers volume discounts when you commit to a minimum monthly spend or annual contract. If you can forecast your call volume with confidence, ask for a tiered rate structure that lowers the per-minute cost as you cross volume thresholds. Typical discounts range from 10–30% off the Developer plan rate for commitments above 10,000 minutes per month, though Vapi does not publish these tiers.

Provide historical call data or a conservative forecast to anchor the negotiation.

Ask what's included in the per-minute rate

Confirm whether the quoted per-minute rate includes telephony, STT, LLM, and TTS, or whether any of those components are billed separately. Ask whether international calls, toll-free numbers, or premium voices carry surcharges. If you plan to use a specific LLM (e.g., Claude 3.5 Sonnet) or voice provider (e.g., ElevenLabs Turbo), request a per-minute rate for that exact configuration rather than accepting a baseline quote that may not reflect your production setup.

Cap overage charges

If you commit to a monthly minimum, negotiate a cap on overage charges to protect against unexpected spikes in call volume. A common structure is to pay the contracted rate up to the minimum, then a slightly higher rate (e.g., 10–20% above the contracted rate) for usage beyond the minimum, with a hard cap at 150–200% of the contracted rate.

This prevents a single high-volume month from doubling your bill.

Negotiate concurrency limits

If your use case requires high concurrency (100+ simultaneous calls), confirm that Vapi's infrastructure can support your peak load and ask whether concurrency limits are enforced. Some voice platforms charge extra for guaranteed concurrency or reserve capacity. If Vapi does not publish concurrency pricing, ask whether your Enterprise contract includes a concurrency SLA and what happens if you exceed it (throttling, additional fees, or best-effort handling).

Pilot before annual commitment

Request a 30- or 60-day pilot period at the contracted per-minute rate before signing an annual agreement. Use the pilot to validate that Vapi's latency, voice quality, and integration capabilities meet your requirements under production load. Negotiate an exit clause that allows you to cancel the annual contract if the pilot does not meet agreed performance benchmarks (e.g., 95% call-completion rate, sub-1-second response time).

Confirm data retention and compliance terms

If your use case requires long-term call recording or falls under regulatory requirements (GDPR, HIPAA, TCPA), confirm that Vapi's data retention and storage policies meet your obligations. Ask whether call recordings and transcripts are stored indefinitely or purged after a set period, and whether extended retention incurs additional fees.

Request documentation of Vapi's security certifications (SOC 2, ISO 27001) and data-processing agreements if you handle sensitive customer data.

Verdict

Vapi's per-minute pricing suits developer teams that want full control over call logic and model selection without the overhead of managing separate STT, TTS, and telephony vendors. The $0.05 baseline rate is competitive for basic configurations, and the pay-as-you-go Developer plan removes the friction of upfront platform fees.

Teams running a few thousand calls per month with standard LLMs and voices will find Vapi straightforward and cost-effective.

The pricing becomes harder to justify at high volumes or when premium models are required. A contact centre running 10,000+ calls per month at $0.30 per minute will pay $24,000+ monthly, which is comparable to hiring human agents in some markets. At that scale, Enterprise pricing with volume discounts is necessary, but Vapi does not publish those rates, so buyers cannot model costs without a sales conversation.

Non-technical teams or those needing pre-built agent templates should compare alternatives with no-code builders or out-of-the-box integrations.

Vapi is the stronger fit for engineering-led startups and mid-market companies building custom voice agents for inbound support, outbound sales, or appointment setting. It is less suitable for teams without in-house developers, buyers needing a unified platform for inbound voice and outbound prospecting, or contact centres that require transparent volume pricing before engaging sales.

Frequently asked questions

How much does Vapi cost per minute?

Vapi's Developer plan starts at $0.05 per minute for basic configurations (standard LLM and voice), with costs rising to $0.15–$0.30+ per minute when using premium models like GPT-4 or premium voices like ElevenLabs Turbo. Enterprise customers negotiate custom per-minute rates based on committed volume.

The per-minute rate includes platform access, telephony, speech-to-text, LLM inference, and text-to-speech, as listed on Vapi's pricing page at time of writing.

Does Vapi have a free trial?

Vapi provides $10 in free credits on signup, which allows new users to test the platform and run a few dozen calls before adding funds. There is no time limit on the credits, but they are consumed as you use the platform.

After the credits are exhausted, you must add funds to continue using the Developer plan or contact sales for Enterprise pricing.

Is telephony included in Vapi's pricing?

Yes, telephony (SIP trunking for inbound and outbound calls) is included in the per-minute rate, as listed on Vapi's pricing page. Phone numbers are billed separately at approximately $1–$2 per number per month for local numbers and $2–$5 per month for toll-free numbers.

The pricing page does not specify whether international calls carry surcharges; buyers should confirm with Vapi's support before launching campaigns to non-US destinations.

Does Vapi charge for failed or unanswered calls?

Vapi's pricing page does not explicitly state whether failed calls (calls that do not connect) or time spent in voicemail detection are metered. Third-party user discussions suggest that Vapi does not charge for unsuccessful dial attempts, but this is not confirmed in official documentation.

Buyers running high-volume outbound campaigns should verify with Vapi's support how failed calls and voicemail detection are billed before committing to a contract.

Is there a minimum contract?

The Developer plan has no minimum contract or spend requirement; it is pay-as-you-go with usage billed in real time. The Enterprise plan typically involves a minimum monthly spend or annual contract, though Vapi does not publish those terms.

Buyers should expect to commit to a volume forecast or minimum spend to access Enterprise volume discounts and dedicated support.

How does Vapi pricing compare to hiring a human receptionist or SDR?

A human receptionist in the US costs $30,000–$45,000 annually in salary plus benefits, which translates to $2,500–$3,750 per month. A Vapi agent handling 500 calls per month at 3 minutes each with a $0.05 per-minute rate costs approximately $75 per month, or 2–3% of a human receptionist's cost.

The comparison shifts at higher volumes or with premium models: 10,000 calls per month at 8 minutes each with a $0.30 per-minute rate costs $24,000 per month, which is comparable to hiring a small team of human agents. Use the SDR cost calculator to model the trade-off for outbound sales use cases.

What's the cheapest alternative to Vapi?

Retell AI and Bland AI both publish per-minute rates in the $0.08–$0.09 range for basic configurations, which is slightly higher than Vapi's $0.05 baseline but comparable when factoring in model and voice selection. Synthflow charges $29 per month plus $0.10 per minute, which is more expensive per minute but may have a lower total cost of ownership for small teams that lack engineering resources and need a no-code builder.

The cheapest option depends on your call volume, model requirements, and whether you have in-house developers.

How does Vapi compare to 11x's Julian?

Julian is an AI phone agent for inbound sales, sold as part of the 11x platform alongside Alice (AI SDR for outbound email and LinkedIn). Pricing is custom and requires a demo. Julian is built for speed-to-lead (answering inbound calls in under 10 seconds) and qualification, with native integration to 11x's outbound engine and 400M+ verified contact database.

Vapi is a developer platform for building custom voice agents across any use case (support, sales, scheduling). Vapi suits engineering teams that want full control over call logic and model selection; Julian suits revenue teams that want a unified platform for inbound voice and outbound prospecting without managing separate tools.

What happens if I exceed my expected call volume?

On the Developer plan, there is no volume cap; you simply pay the per-minute rate for all usage. If you run out of account funds, Vapi will pause service until you add more. On the Enterprise plan, overage handling depends on your contract terms. Some agreements charge a higher per-minute rate for usage beyond the committed minimum; others allow flexible overages at the contracted rate. Negotiate overage terms and caps before signing an annual contract to avoid surprise bills.

Can I switch LLMs or voices mid-month?

Yes. Vapi allows you to change LLM providers and voice engines at any time through the dashboard or API. The per-minute rate adjusts based on the configuration used for each individual call. If you run 500 calls with GPT-3.5 at $0.05/min and 500 calls with GPT-4 at $0.15/min in the same month, you will be billed separately for each rate tier. This flexibility is useful for A/B testing or adjusting quality vs. cost trade-offs as your use case evolves.

Last updated: October 2026.