Retell AI is a platform for building AI voice agents and AI chat agents. It is aimed at teams that want to automate customer calls, sales qualification, appointment workflows, support routing, and other voice-first business interactions.
The official pricing page positions Retell as “true pay as you go”: start at $0, get free credits, and pay only for usage before moving to custom enterprise terms. In June 2026, the important buyer detail is that Retell also exposes component pricing for LLMs, text-to-speech, telephony, speech-to-speech options, active-call concurrency, knowledge-base subscriptions, and call-agent add-ons.
System Verdict
Pick Retell AI when the workflow is a real phone call, not just a chatbot with voice. Retell is strongest when you need call analytics, transcripts, simulation testing, webhooks, API access, concurrency, and enterprise escalation.
Skip it if you want simple fixed-price automation. Voice agents are usage-based systems, and real cost depends on minutes, model choice, speech stack, telephony, and call complexity.
Who pays which tier: Pay-as-you-go is right for builders and pilots. Enterprise is for larger deployments that need SSO, custom agreements, dedicated infrastructure, and higher concurrency.
Key Facts
| Core product | AI voice agent and AI chat agent platform |
| Pricing model | Pay as you go plus Enterprise |
| Free credits | $10 free credits |
| AI voice price | $0.07-$0.31/minute |
| AI chat price | $0.002+/message |
| Concurrent calls | 20 included on pay-as-you-go; extra active-call concurrency listed at $8/month each |
| Developer features | Webhooks and API access |
| Operations features | Call analytics, transcripts, simulation testing, knowledge bases, batch calls |
| Cost add-ons | Knowledge Base, Batch Call, Branded Call, Advanced Denoising, Safety Guardrails, PII Removal, AI Quality Assurance, extra active-call concurrency |
| Enterprise controls | Role-based access control, SSO, custom MSA/DPA/BAA, dedicated stable server, no cap on concurrent calls |
What It Actually Is
Retell is voice-agent infrastructure, not a generic transcription app. The pricing page lists pre-built agent templates, call analytics, transcripts, simulation testing, webhooks, API access, and 20 included concurrent calls on pay-as-you-go. Additional active-call concurrency is listed at $8 per concurrency per month, so busy outbound campaigns should model concurrency separately from per-minute voice cost.
That combination is important. Good voice agents need more than a nice voice. They need interruption handling, fallback behavior, validation, call transfer logic, latency management, transcripts, testing, and post-call data flows. Retell is built around that operational layer.
Retell’s docs also show a knowledge-base layer for website URLs, uploaded documents, and custom text, with automatic retrieval during calls. For outbound work, its batch-call docs support CSV uploads, dynamic variables, scheduling, and status metrics, but calls still start as concurrency slots become available.
Decision Matrix
| Need | Retell AI fit | Notes |
|---|---|---|
| AI phone receptionist | Strong | Templates and call infrastructure help |
| Sales qualification calls | Strong | Works when prompt, CRM, and escalation are designed well |
| Support call automation | Strong | Analytics and transcripts matter |
| Simple voice notes | Weak | Use Wispr Flow or Whisper-based tools |
| Flat monthly cost | Weak | Usage-based voice costs vary |
| Regulated call workflows | Medium | Enterprise terms, BAA availability, PII removal, recording/transcription settings, and retention settings need review |
When To Pick Retell AI
- You need phone call automation. Retell is made for voice agents that talk with customers, not just internal dictation.
- You need developer control. Webhooks and API access are key for production workflows.
- You want to test before committing. Pay-as-you-go plus free credits lowers the first deployment barrier.
- You need observability. Analytics, transcripts, simulation testing, post-call analysis, and QA options are essential for debugging voice agents.
- You need call-time context. Retell knowledge bases can ingest URLs, documents, and custom text, then retrieve relevant chunks during a call.
- You need enterprise escalation later. Dedicated infrastructure, SSO, custom contracts, role-based access control, and no-cap concurrency are available on Enterprise.
When To Pick Something Else
- Voice dictation: Wispr Flow or Whisper.
- Voice model/API building blocks: ElevenLabs, Deepgram, Cartesia, or Hume AI.
- Conversation design and bots: Voiceflow.
- Meeting notes: Granola or Fathom.
Pricing
Retell pricing was verified from the official pricing page on 2026-06-12:
| Plan | Price | Practical meaning |
|---|---|---|
| Pay as you go | $0.07-$0.31/minute for AI Voice Agents | Build and test without an annual contract; 20 concurrent calls included |
| AI Chat Agents | $0.002+/message | Text/chat agent usage, with model-specific per-message pricing |
| Extra concurrency | $8/concurrency/month after the included 20 | Active-call capacity is a separate scaling line item |
| Add-ons | From +$0.005/minute, +$0.005/dial, or fixed monthly line items | Knowledge Base, Batch Call, Branded Call, Safety Guardrails, Advanced Denoising, PII Removal, and AI Quality Assurance can increase the real cost |
| Enterprise | Custom | Dedicated support, custom agreements, SSO, role-based access control, dedicated stable server, and high/no-cap concurrency |
Retell also lists $10 in free credits and 20 included concurrent calls on pay-as-you-go. The real bill depends on call volume, LLM provider, telephony setup, add-ons, knowledge-base count, active-call capacity, and agent configuration. Retell says failed calls are not billed, and voicemail is billed only while the AI agent is active on the line. Retell also says AI voice billing continues during silence or hold time because speech-to-text remains active, but the AI voice agent fee stops after a call is transferred and only telephony continues for the transferred portion.
Against The Alternatives
| Retell AI | Voiceflow | ElevenLabs | |
|---|---|---|---|
| Primary job | AI phone agents | Conversation design and agent workflows | Voice generation and audio AI |
| Pricing unit | Minutes/messages | Seats/usage depending plan | Credits/API usage |
| Developer controls | Strong API/webhooks | Strong design tooling | Strong audio APIs |
| Best for | Calls, call ops, batch calling, and phone-agent context | Bot/agent design | Voice quality and speech |
| Best viewed as | Voice-agent platform | Conversation builder | Audio model platform |
Failure Modes
- Cost forecasting is hard. Minutes, models, TTS, STT, telephony, add-ons, silence time, transfers, and retries all affect real cost.
- Voice agents need narrow workflow design. Broad prompts can fail in live calls.
- Knowledge bases are not magic grounding. Retell retrieves chunks automatically, but the docs still recommend clear source formatting and prompt constraints when the agent must avoid unsupported answers.
- API migrations can break older integrations. clients should migrate before relying on the current dashboard and SDK surface.
- Compliance is not automatic. Healthcare, legal, finance, and HR use cases need careful review of BAA terms, PII removal, recording/transcription opt-outs, and retention settings.
- Human fallback still matters. Agents should transfer or escalate when confidence drops.
- Latency and interruption handling can make or break the experience. Test with real accents, noisy rooms, and impatient callers.
Methodology
This page was produced by the aipedia.wiki editorial pipeline. Scoring follows the four-dimension rubric at /about/scoring/ (Utility x Value x Moat x Longevity, unweighted average). Last verified 2026-06-12 against Retell AI primary pricing, documentation, knowledge-base, batch-call, and API-deprecation information.
FAQ
What is Retell AI? Retell AI is a platform for building AI voice agents and AI chat agents, especially for business phone workflows.
How much does Retell AI cost? The official pricing page lists AI Voice Agents at $0.07-$0.31 per minute, AI Chat Agents at $0.002+ per message, and Enterprise custom pricing. Production costs can be higher when you add specific LLMs, premium voices, telephony, Knowledge Base, Batch Call, Safety Guardrails, PII Removal, or AI Quality Assurance.
Does Retell AI have free credits? Yes. The pricing page lists $10 in free credits.
Does Retell AI support knowledge bases? Yes. Retell’s docs say knowledge bases can use website URLs, documents, and custom text, then retrieve relevant chunks automatically during calls. That helps support and FAQ agents, but the source material still needs clean formatting and testing.
Is Retell AI good for phone support automation? Yes, if the workflow is designed tightly and tested. It includes templates, analytics, transcripts, simulation testing, knowledge bases, webhooks, API access, and batch-call workflows.
Sources
- Retell AI pricing: pay-as-you-go prices, free credits, concurrent calls, features, Enterprise controls
- Retell AI knowledge base docs: URL, document, custom text, retrieval, limits, and best-practice guidance
- Retell AI batch-call docs: CSV upload, dynamic variables, scheduling, status, and concurrency-slot behavior
- Retell AI June 15, 2026 deprecation notice: legacy list endpoint migrations and post-call/post-chat analysis field deprecations
Related
- Category: AI Voice · AI Automation
- Alternatives: Voiceflow · ElevenLabs · Deepgram · Cartesia · Hume AI
