Retell pricing explained: adding up the per-minute components

Retell AI advertises voice agents at "$0.07–$0.31 per minute" — a range, not a price, because Retell's pricing is component-based: you assemble a per-minute rate from a base infrastructure fee plus the language model, voice, and telephony you choose (Retell pricing page, accessed 2026-08-19). The good news for buyers is that, unlike some competitors, every component carries a published number, so you can compute your exact per-minute cost before signing up. This guide walks through the components, adds them up for a realistic workload, and flags the line items that are easy to miss.

How Retell's pricing model works

Retell is pay-as-you-go with no mandatory subscription; new accounts start with $10 in free credits (Retell pricing, accessed 2026-08-19). A connected minute is billed as a stack of components:

1. Retell Voice Infra — $0.055/minute. The base platform fee every voice call pays.
2. LLM — priced per model, per minute. As of 2026-08-19 the published rates run from GPT 5 nano at $0.003/minute up to GPT 5.5 at $0.16/minute — and Retell now also lists "fast" variants at 1.5–2× the standard rate (GPT 5.5 fast is $0.32/minute, GPT 5.4 fast $0.16, GPT 4.1 fast $0.0675). Mid-range options include GPT 4.1 at $0.045, GPT 5.2 at $0.056, and Claude 5 Sonnet and GPT 5.4 at $0.08.
3. Text-to-speech — priced per provider, per minute. Retell platform voices, MiniMax, Fish, Cartesia, and OpenAI voices are $0.015/minute; ElevenLabs voices are $0.040/minute.
4. Telephony — $0.015/minute for US calls via Twilio (rates vary by country), or no charge for SIP trunking if you bring your own telephony. Retell-provisioned phone numbers are $2.00/month each.

Concurrency: 20 concurrent calls are included free; additional concurrency is $8.00 per concurrent line per month. Enterprise plans offer custom pricing, volume discounts, and no concurrency cap (Retell pricing, accessed 2026-08-19; concurrency and enterprise terms also recorded 2026-08-05 in our research and unchanged since).

What's included vs. billed separately

Source: retellai.com/pricing, accessed 2026-08-19.

Worked example: 10,000 minutes, computed from published numbers

Take a common small-business workload: 2,000 calls × 5 connected minutes = 10,000 connected minutes per month, peaking under 20 simultaneous calls (so concurrency is included), one Retell phone number, US telephony, no add-ons. All arithmetic is illustrative, using the rates published on 2026-08-19.

Mid-range configuration — GPT 4.1 + a platform voice:

Voice infra $0.055 + LLM $0.045 + TTS $0.015 + telephony $0.015 = $0.13/minute 10,000 minutes × $0.13 = $1,300, plus $2 for the number = $1,302/month

Budget configuration — GPT 5 nano + a platform voice:

$0.055 + $0.003 + $0.015 + $0.015 = $0.088/minute → $880 + $2 = $882/month

Premium configuration — GPT 5.5 (standard) + ElevenLabs voice:

$0.055 + $0.16 + $0.040 + $0.015 = $0.27/minute → $2,700 + $2 = $2,702/month

Credit where due: the total is fully computable from Retell's public pages — every component has a published number, which is more than can be said for platforms that pass provider costs through unpriced. The spread is the finding here: the same 10,000 minutes lands anywhere from roughly $880 to $2,700+ per month depending on model and voice choices — and a fast-tier build computes higher still: GPT 5.5 fast + ElevenLabs voice + US telephony is $0.055 + $0.32 + $0.040 + $0.015 = $0.43/minute, or $4,302/month with one number. Note the advertised range's two ends: the $0.07 floor is only reachable with the cheapest model and no Twilio telephony (SIP trunking at no charge), and the $0.31 ceiling reflects standard model tiers — fast-tier configurations can exceed it.

Gotcha checklist

• The range is wide — anchor on your configuration, not the floor. "$0.07–$0.31/min" spans more than 4x. Price the specific LLM and voice you intend to run; the floor assumes the $0.003/minute model.
• "Fast" model variants raise the LLM line 1.5–2×. The pricing page lists fast tiers above the standard per-minute rate (GPT 5.5 fast $0.32/minute vs. $0.16 standard; GPT 4.1 fast $0.0675 vs. $0.045). If latency pushes you to a fast tier, your largest component grows by up to double — and the total can pass the advertised $0.31 ceiling.
• ElevenLabs voices cost 2.7x other voices. $0.040/minute vs. $0.015 — on 10,000 minutes that voice choice alone is a $250/month difference.
• Add-ons stack per minute. Knowledge base, denoising, and guardrails at +$0.005 each, PII removal at +$0.01 — enabling all four adds $0.025/minute ($250 on this workload). AI Quality Assurance is the big one: $0.10/minute after 100 free minutes, which can approach or exceed the cost of the underlying call.
• Telephony varies by country. $0.015/minute is the US Twilio rate; international rates differ, and bringing your own SIP trunk removes the telephony line entirely.
• Concurrency is a monthly fixed cost past 20. Bursty inbound traffic that peaks at 40 simultaneous calls means 20 extra lines at $8/month = $160/month even in quiet months.

How ThunderPhone prices differ

ThunderPhone publishes one all-in per-minute engine rate — speech recognition, language model, and voice included — pay as you go with no subscription or seat fee: Spark 2¢/minute, Bolt 5¢/minute, Storm 9¢/minute. The same 10,000-minute scenario computes directly to $200, $500, or $900 per month by engine (illustrative arithmetic; premium voice or language paths can add up to 3¢/minute, hold time bills a flat 2¢/minute, and telephony depends on setup). There is no separate model-selection line to price: the engine rate is the conversation rate. ThunderPhone is GDPR and HIPAA compliant, with customers remaining responsible for configuring their workflows, data handling, recordings, and notices for their own obligations.

FAQ

How much does Retell AI cost per minute?

Retell advertises $0.07–$0.31 per minute, assembled from components — and a fast-tier configuration can compute above that ceiling. Every call pays $0.055/minute voice infra, plus your LLM ($0.003–$0.16/minute for standard models, up to $0.32/minute for fast variants), plus TTS ($0.015 or $0.040/minute), plus telephony ($0.015/minute US, or no charge via SIP trunking). Retell pricing, accessed 2026-08-19.

Does Retell charge a monthly subscription?

No mandatory subscription for pay-as-you-go use, and new accounts get $10 in free credits. Monthly fixed costs only appear for phone numbers ($2.00/month, $10.00/month verified), extra concurrency beyond the included 20 calls ($8.00/concurrency/month), and knowledge bases beyond ten ($8.00/KB/month). Enterprise plans are custom-priced. Retell pricing, accessed 2026-08-19.

What would 10,000 minutes a month cost on Retell?

Roughly $880 to $2,700+ depending on configuration, computed from published rates — with fast-tier builds higher still. A GPT 4.1 + platform-voice + US-telephony setup works out to $0.13/minute ($1,300); the cheapest published stack is about $0.088/minute ($880); a GPT 5.5 + ElevenLabs setup about $0.27/minute ($2,700); and a GPT 5.5 fast + ElevenLabs setup about $0.43/minute ($4,302 with one number), all illustrative. For contrast, ThunderPhone's all-in rates put the same workload at $200/$500/$900 by engine.

Does Retell charge for telephony and phone numbers?

Telephony yes, SIP no. US telephony via Twilio is $0.015/minute (varies by country), while SIP trunking with your own telephony carries no Retell charge. Retell-provisioned numbers are $2.00/month; verified numbers are $10.00/month. Retell pricing, accessed 2026-08-19.

Is Retell HIPAA and GDPR compliant?

Retell's documentation states SOC 2 Type 1 and Type 2, HIPAA compliance with a signed BAA required before transmitting PHI, and GDPR compliance via a Data Processing Addendum with SCCs on AWS infrastructure. The same page notes Retell does "not currently operate services within the European Union." (Retell compliance docs, accessed 2026-08-18.)

Sources and freshness

All Retell claims on this page were checked against the linked first-party pages on the accessed dates shown beside each claim. ThunderPhone pricing and product details reflect its published pricing and documentation as of August 19, 2026. This guide is documentation-based — no Retell account, invoice, or live call was tested — and vendors change pricing frequently, so confirm against Retell's live pages before relying on them.