Cekura has raised $2.4M to help make conversational agents reliable

Retell AI Pricing per Minute: What You Actually Pay in 2026

Team Cekura

Written by:

Team Cekura

Shashij Gupta

Reviewed by:

Shashij Gupta

Last updated

May 19, 2026 · 10 min read

Retell AI pricing per minute starts at $0.07, but it doesn't include the LLM, telephony, or anything else you actually need to run a call. Add those in, and most setups land between $0.13 and $0.31/min. Here's the full breakdown so you know what to expect in costs.

Retell AI Pricing Plans: At a Glance

There are two tiers: Pay-as-you-go for building and testing, and Enterprise for running production calls at volume and requiring dedicated infrastructure, higher concurrency caps, and compliance documentation.

PlanPriceBest ForKey Features
Pay-as-you-go$0.07-$0.31/minDevelopers building and testing$10 free credits, 20 concurrent calls, 10 free Knowledge Bases
EnterpriseCustomHigh-volume teamsDedicated server, no concurrent call cap, HIPAA/BAA, 24/7 support

Understanding Retell AI pricing per minute requires tracking each component separately. Voice infra plus standard TTS gets you to $0.07/min. Stack an LLM and telephony on top, and the minimum jumps to $0.088/min.

How Retell AI Pricing Works: Pay-As-You-Go

ComponentCostWhat It DoesNotes
Voice Infra$0.055/minCore voice layer, required for all agentsBase of every setup
Text-to-Speech$0.015/minStandard voices (Retell, Minimax, Cartesia, OpenAI)ElevenLabs costs $0.040/min
LLM$0.003-$0.160/minUnderstands and responds to the callerGPT 5 nano ($0.003) to GPT 5.4 Fast ($0.160)
Telephony$0.015/minConnects the agent to the phone networkUS rate. SIP/custom telephony is free
Knowledge Base$0.005/minThe agent reads documents during callsFirst 10 free, $8/month each after
Phone Numbers$2.00/monthDedicated number for your agentOptional. Own numbers accepted at no charge

Retell AI Pricing Plans Breakdown

Two plans, one pricing model. Here's what Retell AI pricing per minute looks like inside each one.

Pay-As-You-Go

What's included:

  • $10 in free credits to start
  • Up to 20 concurrent calls (more available on demand)
  • 10 free Knowledge Bases (additional at $8.00/month each)
  • All voice models, including Retell native, ElevenLabs, OpenAI Voices
  • All LLM options like GPT 5 nano, GPT 4.1 Mini, Claude 4.5 Haiku, Claude 4.6 Sonnet, and more
  • Retell phone numbers at $2.00/month, or bring your own
  • SIP Trunking and custom telephony at no additional charge

Best for: Developers building and testing, and startups with unpredictable call volumes.

Pros:

  • No monthly commitment. You only pay for actual usage.
  • You pick the LLM and voice model, so you control what each call costs.
  • $10 in free credits to test before spending anything.

Cons:

  • 20 concurrent call limit can become a bottleneck for growing operations.
  • Per-minute costs add up across components. Voice infra, TTS, LLM, and telephony each have their own rate, so the $0.07/min floor doesn't reflect what most setups actually pay.

Enterprise: Custom Pricing

What's included:

  • Dedicated stable server
  • No cap on concurrent calls
  • HIPAA/BAA compliance
  • 24/7 support with dedicated portal
  • Custom telephony and integrations

Best for: High-volume operations and teams in regulated industries like healthcare or finance that need HIPAA coverage and dedicated infrastructure.

Pros:

  • No cap on concurrent calls, so volume spikes don't become a billing problem.
  • HIPAA/BAA coverage makes it viable for regulated industries.
  • Dedicated server means your calls aren't competing for resources with other customers.

Cons:

  • Pricing isn't public. You need to contact sales before you can forecast your costs.
  • Retell doesn't publish minimum contract terms or volume commitments. Verify with their sales team before signing.

Which Retell AI Plan Should You Choose?

Retell AI bills every component separately. Voice infra, TTS, LLM, and telephony each run on their own meter.

That structure means the plan choice comes down to whether you can absorb variable per-minute stacking costs or need predictable enterprise-level infrastructure.

Choose pay-as-you-go if you:

  • You're still selecting your LLM stack. The plan lets you test GPT-5 Nano ($0.003/min) against Claude 4.6 Sonnet ($0.08/min) without locking into a configuration.
  • Run under a limited number of concurrent calls consistently. Retell's cap on this tier is a hard ceiling with no grace period.
  • Want to use your own SIP trunk to eliminate the $0.015/min telephony charge.
  • Are prototyping with the 10 free Knowledge Bases before deciding if the $8.00/month per-base cost scales for your use case.

Choose enterprise if you:

  • Need concurrent call volume beyond a limited number. The pay-as-you-go cap is a hard limit, and the only way past it is Enterprise.
  • Are building for healthcare and need a signed BAA. The pay-as-you-go plan doesn't include HIPAA/BAA coverage.
  • Require a dedicated server to avoid latency spikes. Shared infrastructure causes latency spikes that show up in production response times.
  • Need cost predictability at high volume. Per-minute stacking across all components makes pay-as-you-go increasingly unpredictable as call minutes scale.

Is Retell AI Worth the Cost?

It depends on who's running it. The platform gives you real infrastructure control, but that control requires active management. If no one on your team is watching component costs, the bill will surprise you.

One cost the per-minute pricing breakdown doesn't capture is validating that your agent handles real callers correctly before you pay for thousands of live minutes. That testing layer is separate from the infrastructure itself.

Retell AI is worth it if you:

  • Need granular control over your LLM stack. You can mix a cheaper model for simple flows and a stronger one for complex reasoning on the same platform.
  • Are building production voice agents that need API-first architecture and SIP trunk support. Most no-code tools cap out before Retell does.

Skip Retell AI if you:

  • Need a fixed monthly cost from day one. Forecasting is difficult until you have real call data to model against.
  • Don't have a developer available to manage the stack. The integration work and ongoing maintenance add time and cost that don't appear on the invoice.
  • Need to validate agent performance before scaling. Retell AI pricing covers the infrastructure, but many teams have to add another tool on top to test agent behavior across scenarios before committing to production-scale call volume.

Retell AI Alternatives & Pricing Comparison

Not every voice AI platform bills the same way. Here's where Retell lands and how it compares to the main options.

ToolStarting PriceBest ForKey Advantage
Retell AI$0.07-$0.31/min (Pay-as-you-go)Developers who need full LLM and voice model controlComponent billing is flexible. Mixes GPT 5 nano vs. Claude 4.6 Sonnet per use case
Vapi AI$0.05/min (Vapi Hosting) + model providers at costEngineering teams deploying voice agents at scale10 concurrent lines included + Custom SIP + SMS/Chat on the same PAYG plan
Bland AI$0.14/min (all included)Teams that want a flat rate with no component stackingLLM + STT + TTS + telephony all in one per-minute rate
Synthflow AI$0.09/min (Voice Engine) + LLM separateBuilders and agencies needing white-label and resellerWhite-label toolkit + SOC2, GDPR, ISO 27001 on PAYG

Cekura + Retell AI: Testing and Monitoring Your Voice Agent

Retell AI runs the agent. Cekura makes sure it actually works before it goes live and keeps tabs on it after. They're not competing tools. They're two parts of the same setup.

The infrastructure, TTS, LLM routing, and telephony all operate on the platform side. From there, a single webhook pointing to Cekura's endpoint connects the monitoring layer.

When a call ends, it pulls the audio and transcript automatically and runs quality checks across the full conversation, logging instruction-following, tool usage, latency, and hallucination detection, before and after you go live.

Use Retell AI when you need:

  • You need to choose a different LLM or voice model depending on the use case.
  • API-first infrastructure with SIP trunk support for production deployments.
  • You want to test different configurations on pay-as-you-go before locking into anything.

Use Cekura when you need:

  • Pre-production simulation of your Retell AI agent across hundreds of scenarios.
  • Production monitoring, including alerts, call analytics, and quality metrics on live conversations.
  • Automated QA for your CI/CD pipeline, so test runs trigger automatically on prompt changes.
  • Predefined and custom metrics, with built-in latency, instruction-following, and tool-call tracking, plus define your own.
  • Custom personas and accents that you can test against Cekura's curated caller library, or build your own with specific accents and background noise.
  • Customer satisfaction metrics to track CSAT and drop-off points to find where your agent loses callers.
  • SOC 2, HIPAA, and GDPR compliance for transcript redaction, role-based access, and audit trails.

Cekura also integrates natively with the platforms most teams already run. Native integrations work out of the box for Retell, VAPI, ElevenLabs, LiveKit, Pipecat, Synthflow, Bland, Cisco, and more.

Use both, and you've got the build and the QA covered. Book a demo to see how Cekura monitors your Retell AI agent.

My Bottom Line on Retell AI Pricing

Retell AI pricing per minute works if you have someone watching the component stack. What catches most teams off guard is the gap between the $0.07/min advertised rate and the actual rate they pay once LLM and telephony kick in.

Start on pay-as-you-go and track your per-component costs from day one. When you're ready for production, Cekura is the layer that tells you whether your agent actually works in production.

Frequently Asked Questions

How Much Is Retell AI Pricing per Minute?

Retell AI pricing per minute starts at $0.07, but that only covers the voice infrastructure layer. A typical setup with voice infrastructure, TTS, LLMs, and telephony costs between $0.13 and $0.31/min, depending on the models you choose.

Does Retell AI Charge for Failed Calls?

No, Retell AI doesn't charge for calls that fail to connect. Voicemail is the exception: billing applies only for the time the agent is active on the line.

What Is the Difference Between Retell AI and Vapi?

The main difference between Retell AI and Vapi lies in their platform fee structures. Retell charges $0.055/min for voice infra plus component costs, while Vapi charges $0.05/min for hosting and passes provider costs through at cost.

Both follow modular billing, but Retell includes 20 concurrent calls by default, while Vapi includes 10.

Does Retell AI Offer a Free Trial?

Yes, new accounts get $10 in free credits. At a typical rate of $0.11-$0.15/min, that covers roughly 67-90 minutes of testing before any payment is required.

Is Retell AI HIPAA Compliant?

Yes, Retell AI is HIPAA compliant, but only on the Enterprise plan. The pay-as-you-go plan doesn't include a BAA, so healthcare teams need to contact sales to arrange compliance coverage.

Ready to ship voice
agents fast? 

Book a demo