Vapi
Developer platform to build, test, and deploy AI voice agents for phone calls with usage-based per-minute pricing
Pay-as-you-go platform for building AI phone agents that handle calls, transfers, and bookings
Retell AI builds voice agents that answer and place phone calls, transfer to humans, navigate IVR menus, and book appointments. Pricing is genuinely pay-as-you-go with no annual contract: $0.07 to $0.31 per minute for voice agents and from $0.002 per message for chat agents, with $10 in free credits and 20 concurrent calls included. The per-minute bill stacks voice infrastructure at $0.055, text-to-speech, and the LLM you pick.
Retell AI is a platform for building voice agents that work over the phone. An agent can answer inbound calls, place outbound ones, transfer to a human mid-conversation, check availability and book appointments, ground its answers in a knowledge base, and even navigate another company’s keypad IVR menu. Operational tooling sits alongside it: batch calling for campaigns, branded caller ID and verified numbers to avoid spam flagging, post-call analysis, live call monitoring, custom dashboards, and simulation testing before an agent goes live. Telephony arrives through Twilio, Vonage, or a custom carrier, and the platform integrates with Make, n8n, HubSpot, and Go High Level.
The pricing position is the differentiator. Retell leads with the claim that most voice AI platforms require annual contracts before you write any code, and instead publishes a per-component rate card: voice infrastructure at $0.055 per minute, text-to-speech from $0.015 per minute — $0.040 if you want ElevenLabs voices — and a language model billed per minute with an optional faster tier at roughly double the rate. Those components add up to the advertised $0.07 to $0.31 per minute range, with telephony charged separately on top. Chat agents are billed from $0.002 per message. New accounts get $10 in free credits and 20 concurrent calls, and there is no permanent free allowance beyond that. The model names on the rate card are the vendor’s own labels at the time of verification.
Retell AI fits teams that want phone automation without a procurement cycle. Because everything is metered per minute, a small business can run a single receptionist agent for a few dollars a month while a contact centre can scale the same setup — though anything regulated pushes you toward an Enterprise agreement.
Starting price: $0.07/min · Free tier: no · Model: usage-based
Price history tracked from June 2026
| Plan | Price | Includes |
|---|---|---|
| Pay as you go | $0.07-$0.31/min | $10 in free credits to start · 20 concurrent calls included · From $0.002 per message for chat agents · Full platform access with no commitment · Prebuilt agent templates, analytics, and transcripts · Simulation testing, webhooks, and API access · Shared infrastructure, community and email support |
| Enterprise | Custom | No cap on concurrent calls · Dedicated stable server with a built-in fallback system · HIPAA and BAA, custom MSA and DPA · Single sign-on and role-based access control · Custom data retention · 24/7 support, dedicated portal, named account manager |
| Pros | Cons |
|---|---|
| True pay-as-you-go with no annual contract, which is unusual in enterprise voice AI | The $0.07 per minute floor only applies with the cheapest model and voice — voice infrastructure alone is $0.055 per minute before any LLM or speech cost |
| Component rates are published individually, so you can calculate a call's cost before building | ElevenLabs voices cost $0.040 per minute against $0.015 for platform, Cartesia, Fish, MiniMax, or OpenAI voices, nearly tripling the speech line |
| 20 concurrent calls are included on the self-serve plan without negotiation | Choosing a fast-tier LLM roughly doubles that component's per-minute rate |
| Model and voice choice are yours — cheaper LLMs and platform voices cut the per-minute rate substantially | Telephony is billed separately on top of the platform rate, so quoted minute costs are not the final bill |
| Simulation testing and PII removal are available without moving to an enterprise contract | HIPAA and BAA coverage, SSO, dedicated servers, and uncapped concurrency all require an Enterprise agreement with no published price |
| Self-serve accounts run on shared infrastructure, with dedicated stable servers reserved for Enterprise |
Developer platform to build, test, and deploy AI voice agents for phone calls with usage-based per-minute pricing
AI voice platform for text-to-speech, voice cloning, and conversational agents in 70+ languages
Low-latency voice AI platform behind the Sonic text-to-speech model, built for real-time agents
Empathic voice AI for developers — Octave 2 expressive text-to-speech and the EVI speech-to-speech API with per-character and per-minute pricing
Developer-first voice AI API for speech-to-text, text-to-speech, and real-time voice agents
Voice agents cost between $0.07 and $0.31 per minute depending on the components you choose. The range comes from stacking Retell's voice infrastructure at $0.055 per minute, a text-to-speech provider, and the language model. Chat agents start from $0.002 per message instead of per minute.
Not a recurring one. New accounts receive $10 in free credits and full platform access, but there is no permanent monthly free allowance. Once the credits are used you pay per minute or per message. There is also no annual contract requirement, so spend stops when usage stops.
Four components: Retell's voice infrastructure at $0.055 per minute, text-to-speech from $0.015 per minute, the language model, and telephony billed separately. Optional add-ons such as knowledge base, advanced denoising, safety guardrails, PII removal, and AI quality assurance can add more.
The pricing page lists platform voices plus Cartesia, Fish, MiniMax, OpenAI, and ElevenLabs for speech, with ElevenLabs the most expensive at $0.040 per minute. For reasoning it lists a range of OpenAI models and Claude Sonnet, each with a published per-minute rate and an optional faster tier.
The self-serve pay-as-you-go plan includes 20 concurrent calls, with more available on request. Enterprise agreements remove the cap entirely and add a dedicated stable server plus a built-in fallback system, which matters for contact-centre volumes where call spikes are routine.
Only under an Enterprise agreement. HIPAA coverage with a BAA, custom data retention, custom MSA and DPA terms, single sign-on, and role-based access control are all listed as Enterprise features. The self-serve tier does include PII redaction and opt-out recording, but not the contractual protections.