Vapi

Overview

Vapi is a developer-first platform for building and operating voice agents, with tools for live call control, telephony, and model selection. It is strongest for teams that want to shape voice workflows around their own systems and integrations. Its breadth is appealing, though call capacity, data retention, and support change substantially by plan.

Key features

Vapi brings orchestration and real-time monitoring together with configurable enterprise controls. It says responses can arrive in under 500ms, which is aimed at keeping live exchanges moving. Turn-taking, human-like voices, and custom pronunciation and keywords give builders ways to tune how agents speak and respond.

Agents can draw on external data and trigger tools, including MCP, during calls. Teams can bring their own model or select from a managed catalog of more than 200 models and integrations. The breadth supports experimentation and custom stacks, though model costs pass through at cost on every plan, separate from Vapi hosting charges.

Telephony includes inbound and outbound calling, custom SIP, phone number management, DTMF, voicemail detection, and warm transfers. That combination suits workflows where agents must manage phone interactions and pass a call to a person when needed. Vapi also displays integrations with OpenAI, Anthropic, Twilio, Zendesk, Salesforce, HubSpot, and Slack.

Automated test suites, analytics dashboards and exports, and A/B experiments for prompts, voices, and flows support ongoing refinement. Agents can converse in English, Spanish, Mandarin, and more than 100 other supported languages. The Trust Center provides compliance resources for CCPA, GDPR, HIPAA, SOC 2, PCI DSS v4.0.1, and SOC 3, making the platform relevant to teams evaluating regulated use cases.

Pricing

The No Success Package starts free, with $5 in credits, two organizations, one included Vapi phone number, four concurrent calls, and 14 days of raw data retention. Vapi hosting costs $0.05/min, while models are billed at pass-through cost. It includes the full model catalog and platform access, Discord peer support, and email support without an SLA. This is a useful small-scale starting point, but four simultaneous calls and a short retention window limit its role for sustained operations.

Core costs $29 a month, billed prepaid with auto-renewal. It raises capacity to five included phone numbers and 10 concurrent calls, with two organizations and 30 days of raw data retention. Zero data retention (ZDR), email support, and a two-business-day response SLA are included. Core is the clearest fit for a team moving beyond the starter allowance, but the prepaid renewal and pass-through model charges still matter.

Pro costs $999 a month, with billing at 10% of Vapi hosting fees and a $999/mo minimum. It includes 10 organizations, 10 phone numbers, 30 concurrent calls, and 180 days of raw data retention, plus role-based access control (RBAC) and ZDR. Dedicated Slack and email support, a one-business-day response SLA, and a 99% uptime SLA make it more appropriate for teams with heavier operational needs. The minimum price makes it a steep step up from Core.

Premier has custom pricing and is sold directly. It offers elevated organization, concurrency, and retention limits, a custom phone number count, priority support, a named support team and onboarding, a one-hour P0 response SLA, and a 99.9% uptime SLA. Custom RBAC, ZDR, SSO, and dedicated SIP are aimed at organizations with specific scale or control requirements; teams without those needs may find Core or Pro more proportionate.

Platforms

Vapi supports Android, iOS, web, and API access. That mix serves teams building across mobile and web channels, as well as developers integrating voice capabilities directly.

Who it's for

Vapi suits developers and businesses building voice agents for inbound or outbound calls, especially when they need model choice, live tool use, telephony controls, testing, and analytics in one platform. Its plan ladder gives growing teams options for more concurrent calls, longer retention, and stronger support. It is less compelling for small projects that need a generous ongoing free allowance or for teams that do not need the configuration and operational controls behind the higher tiers.

Pros and cons

• Broad call workflow coverage: inbound and outbound calls, voicemail detection, DTMF, warm transfers, and custom SIP cover varied phone interactions.
• Choice of model stack: bring a model or choose from more than 200, with external data and live tools available to agents.
• Useful iteration tools: automated tests, analytics exports, and A/B experiments support ongoing improvement.
• Starter capacity is limited: four concurrent calls and 14 days of raw data retention constrain the no-cost package.
• Pro is a major price jump: its $999/mo minimum may be hard to justify for teams that need more than Core but not the expanded limits and service terms.
• Model charges are additional: pass-through model costs apply across all plans, alongside hosting charges where applicable.

Alternatives

Choose LiveKit Agents if its free Build plan's 1,000 agent session minutes, five concurrent sessions, and one deployment suit your initial workload, or if you need its broader listed platform coverage, including Linux, macOS, Windows, and self-hosting.

Rapida AI has a free Launch plan for one agent and lets teams bring their own STT, TTS, LLM, and SIP, with 10-day observability retention and community support; it may suit teams wanting that bring-your-own stack at the start.

Voiceflow is an alternative for agencies and partners: its free trial requires no credit card and includes usage-based billing, multi-client workspace management, and white-labeling.

Yellow.ai offers a free plan with one AI agent, 500/month included, a two-agent limit, 1,000 records, and one active campaign; consider it if those included allowances match the use case.

Intervo is another freemium option, with a self-hosted plan that includes full source code under an MIT License and community support.

Retell AI offers pay-as-you-go pricing of $0.07-$0.31 per minute, $10 in free credits, and 20 concurrent calls; it is worth comparing for teams weighing usage pricing and a larger included call allowance.

Ring AI has a free tier with 100 minutes/month for testing, while its Enterprise plan is billed per minute at $0.10 per contact under multi-year contract pricing.

Tone offers a free self-host plan with unlimited agents, calls, and workflows, and lets teams bring their own LLM, voice, and telephony.

Browse AI Voice Agents or Conversational AI Platforms to compare more options in those categories.

Verdict

Vapi is a strong choice for developer-led teams that need configurable phone agents, broad model options, and tools to test and refine live workflows. Its clearest advantage is the combination of telephony controls, live tool use, and operational features. Look elsewhere if you need substantial free usage or a lower-cost middle tier: the starter allowance is constrained, and Pro begins at a $999/mo minimum.