Vapi Review 2026: Developer-first voice AI agents with a transparent $0.05/min platform fee
Affiliate disclosure: this review contains affiliate links — we may earn a commission if you sign up, at no cost to you. Ratings are our own editorial scores.
Vapi
Pros
- Transparent pricing: $0.05/min platform fee with provider costs passed through at cost, no markup
- Bring your own STT/LLM/TTS API keys and provider fees drop to $0 on Vapi's side
- Huge model flexibility — mix and match transcribers, LLMs, and voices from providers like Deepgram, OpenAI, Anthropic, and ElevenLabs
- $10 free credit at signup with no credit card required
Cons
- Real all-in cost runs $0.13–$0.31+/min once STT (~$0.01), LLM ($0.02–$0.20), TTS (~$0.04), and telephony (~$0.01) are added
- Built for developers — steep learning curve compared to no-code rivals like Synthflow
- Compliance is pricey: HIPAA costs $2,000/mo and Zero Data Retention another $1,000/mo
- Self-serve Build plan support is limited to community Discord and email
Best for: Developers building custom voice agents through an API-first platform, Agencies deploying white-labeled AI phone agents for multiple clients, Startups prototyping inbound/outbound call automation before committing to enterprise contracts.
What is Vapi?
Vapi is a developer platform for building voice AI agents that hold real phone and web conversations. The homepage frames the goal as speaking human to every customer, and the mechanics underneath are straightforward: speech-to-text converts caller audio, a large language model decides what to say next, and text-to-speech renders the reply as natural speech. Vapi orchestrates that loop, manages the telephony layer, and exposes it through an API.
What separates Vapi from packaged voice products is that nearly every component stays swappable. The documentation lists transcription, model, and voice providers including Deepgram, Gladia, OpenAI, Anthropic, Google, and ElevenLabs, while the platform page counts more than 200 models and integrations. Teams can bring their own provider keys, and Vapi passes model provider charges through at cost.
Assistants and Squads for multi-agent orchestration
An Assistant in Vapi is a single system prompt plus tools and structured outputs, which the quickstart pitches at customer support, lead qualification, and booking and routing. When one prompt starts buckling under too many jobs, Squads splits the work into specialized assistants that hand off mid-call through handoff tools while carrying full conversation context across the transfer. Assistant Overrides and Member Overrides change configuration without modifying the saved assistants, while context engineering controls what conversation history is passed between them and variable extraction saves information to carry across a handoff. Worth knowing: Vapi retired its older visual Workflows builder in August 2026 and now steers builders toward Squads, so node-and-edge tutorials you find elsewhere are out of date.
Tools, telephony, and the developer surface
Tools are where a Vapi agent stops talking and starts doing. The dashboard Tools section creates reusable tools that attach to any assistant, and the documentation covers function, apiRequest, code, and handoff types alongside built-ins for transferCall, endCall, DTMF, voicemail, SMS, Slack messages, MCP servers, and Make. On the telephony side, Vapi handles custom SIP, warm transfers, voicemail detection, phone number management, and DTMF input. Free Vapi numbers cover US area codes only, up to five per account. A Vapi CLI brings the same surface to the terminal, and the quickstart promises a working phone agent in under five minutes.
Latency, fallbacks, and testing before launch
Voice is unforgiving about delay, and Vapi leans into that: the documentation cites sub-600ms response times, while the platform page claims responses under 500 milliseconds, a small inconsistency worth testing yourself. More useful for production teams are the safeguards around failure. Model Fallbacks give automatic failover and latency balancing across model providers, and automated testing plus A/B experiments let you test variations of prompts, voices, and flows before real callers hear them. Reporting and analytics, multilingual support, and AI guardrails round out the operational layer.
Who runs Vapi in production
The Vapi customers page reads like a cross-section of call-heavy industries: Amazon Ring, Kavak, ServiceTitan, New York Life, Intuit, MindTickle, Instawork, Fleetworks, Spring Venture Group, and UnityAI, spanning automotive, freight, healthcare, home security, insurance, and marketplace platforms. The vendor reports that Ring selected Vapi after evaluating dozens of vendors and now routes its inbound volume through the platform, that Instawork processes over a million minutes of voice screening each month, and that Fleetworks handles more than ten thousand AI calls daily. Vapi lists SOC 2, HIPAA, and PCI compliance, plus SSO, OAuth, and RBAC.
Who should choose Vapi
Vapi suits engineering teams that want to own the voice stack rather than rent a packaged version of it. If you have an API to call, a model preference, and the appetite to tune transcription, prompts, and voices yourself, the Build plan lets you start small and scale into per-minute economics that stay legible as volume climbs. High-volume operations with compliance obligations land better on Scale, alongside the HIPAA and Zero Data Retention add-ons. Vapi is a weaker fit for non-technical teams wanting a finished agent out of the box, since the platform hands you primitives and an excellent API rather than a done-for-you contact center. Budget for a developer.
Key features
| Feature | What it does |
|---|---|
| Voice pipeline orchestration | Real-time turn-taking engine that coordinates speech-to-text, LLM reasoning, and text-to-speech for low-latency phone conversations |
| Provider marketplace with BYO keys | Swap STT, LLM, and TTS providers freely; bring your own API keys and pay $0 in provider fees through Vapi |
| Telephony and SMS | Phone calling via carrier integrations plus SMS/chat messaging billed at $0.005 per message |
| Call concurrency scaling | 10 concurrent call lines included on Build; additional lines cost $10/line/mo, with custom concurrency on Scale |
| Enterprise compliance stack | SOC 2, PCI, SSO, and RBAC on Scale contracts; HIPAA ($2,000/mo) and Zero Data Retention ($1,000/mo) add-ons on any plan |
| API-first developer tooling | REST API, SDKs, and webhooks for programmatic assistant creation, tool calling, and call analytics |
Vapi pricing
| Plan | Price | Included |
|---|---|---|
| Free Trial | $5 in free credits | |
| Build (Pay-as-you-go) | $0.05/min | |
| Build — SMS/Chat | $0.005/msg | |
| Build — extra call concurrency | 10 included + $10/line/mo | |
| Scale (Enterprise) | Custom | |
| HIPAA add-on | $2,000/mo | |
| Zero Data Retention add-on | $1,000/mo |
How Vapi compares
| Alternative | How it differs |
|---|---|
| Retell AI | Starts at $0.07/min with no platform subscription — simpler pay-as-you-go for low-volume pilots, less model flexibility |
| Bland AI | Repriced December 2025 to ~$0.14/min self-serve; ~$0.11/min all-in on Scale with bundled billing instead of pass-through fees |
| Synthflow | No-code, subscription-based alternative; ~$0.08–$0.09/min headline but real costs of $0.15–$0.37/min after add-ons |
Vapi ratings on other platforms
Independent user ratings from third-party review sites, linked here for transparency. These are not our editorial score, are captured on the date shown, and may have changed since.
Frequently asked questions
How much does Vapi cost per minute?
Vapi's platform fee is $0.05 per minute, but that only covers orchestration. You also pay at-cost provider fees: transcription around $0.01/min, LLM $0.02–$0.20/min, voice generation about $0.04/min, and telephony roughly $0.01/min. Realistic all-in costs land between $0.13 and $0.31 per minute. Bring your own API keys and those provider layers pass through at $0.
Does Vapi have a free plan?
Not a permanent one. New accounts get $10 in free credit — no credit card required — which covers roughly 60+ test minutes depending on your model stack. After that, everything is pay-as-you-go at $0.05/min plus provider costs. A startup grant program offering up to 90,000 free minutes has existed but is currently on hold.
Is Vapi cheaper than Retell AI?
On headline rates it looks cheaper: Vapi charges $0.05/min versus Retell's $0.07/min. But both pass through provider costs, so real-world totals converge around $0.13–$0.31/min. Vapi wins if you bring your own API keys or need deep model customization; Retell's no-platform-fee pay-as-you-go model is simpler for small pilots.
How much does Vapi HIPAA compliance cost?
HIPAA compliance is a $2,000/month add-on available on both Build and Scale plans, and Zero Data Retention costs another $1,000/month. At low call volumes those fees can dwarf usage charges, making Vapi expensive for small healthcare deployments. Enterprise Scale contracts bundle SOC 2, PCI, SSO, and RBAC with committed-volume discounts.
What does Vapi's Scale enterprise plan cost?
Vapi doesn't publish Scale pricing. It's an annual contract built around a fixed platform fee plus committed call volume, which lowers the effective per-minute rate below the $0.05 self-serve price. It adds enterprise compliance (SOC 2, HIPAA, PCI), SSO and RBAC, custom concurrency beyond the default 10 lines, and a dedicated account team.
Verdict
Buy Vapi if you have engineering resources and want maximum control over every layer of the voice stack — its $0.05/min fee with at-cost provider pass-through is the most transparent pricing model in the category, especially with your own API keys. Skip it if you need no-code simplicity, predictable all-in billing, or if $2,000/mo HIPAA fees break a small healthcare budget.
Facts verified against: vapi.ai, www.cekura.ai, www.cloudtalk.io, www.lindy.ai, vapi.ai, vapi.ai, vapi.ai, docs.vapi.ai, docs.vapi.ai, docs.vapi.ai, docs.vapi.ai, docs.vapi.ai (as of August 2026).
More in AI & AI Agents
See all →
AI video-ad generator that turns product URLs into UGC-style ads
Autonomous AI CRM that prospects, enriches, and follows up for founder-led teams
AI answer engine that searches the web and cites its sources