Retell AI pricing

About

Retell AI is a conversational voice-agent platform that lets contact centers and developers deploy AI phone, chat, SMS, and email agents that handle tier-one calls. A one-line integration composes the full real-time voice stack — speech-to-text, a large language model, text-to-speech, and telephony — behind a single API, with pre-built templates, simulation testing, call analytics, and webhooks. Founded in 2023 (Evie Wang, Zexia Zhang, Bing Wu, Todd Li, Weijia Yu) and a Y Combinator W24 company, Retell launched in February 2024 and reached roughly $3M annualized revenue within months. It announced a $4.6M seed led by Alt Capital (with Y Combinator, Carya Venture, and 20+ founders/operators) in September 2024 and has since scaled to about $60M ARR by April 2026 — up ~650% year-over-year — powering 50M+ real-time AI phone calls per month for BPOs and enterprises like Everise, Sunshine Loans, and Matic.

For the most current information, visit Retell AI.

Pricing summary : How Retell AI’s pricing model works

Retell AI is true pay-as-you-go: there is no platform fee, no contract, and no minimum. Voice agents cost $0.07–$0.31 per minute, and the headline figure is unbundled — every minute is itemized into Retell Voice Infra ($0.055/min), the LLM you choose (e.g. GPT 5.1 at $0.04/min standard, GPT 5.5 at $0.16/min, Claude 4.6 Sonnet at $0.08/min), text-to-speech ($0.015/min for most voices, $0.04/min for ElevenLabs), and telephony ($0.015/min via Twilio/Telnyx, or free if you bring your own SIP trunk). Retell’s on-page calculator shows a computed all-in estimate of roughly ~$0.115/min for a default GPT-class agent (see Hidden costs). Chat agents are billed per AI message from $0.002+ (e.g. GPT 5.1 at $0.013/msg). New accounts get $10 in free credits and 20 free concurrent calls to build and launch.

The Enterprise plan is custom-quoted and adds volume pricing, a dedicated stable server, HIPAA/BAA, custom MSA/DPA, custom SSO, role-based access control, a high concurrency cap, and 24/7 omnichannel support with a dedicated portal.

Call-agent pricing has two modes on the pricing page. The Default (cascading) mode is the unbundled stack above. A Speech-to-Speech tab prices realtime speech-to-speech engines as a single per-minute line instead: GPT Realtime 1.5 and GPT Realtime at $0.345/minute, and GPT Realtime mini at $0.07/minute.

What makes this different: Retell exposes the entire cost stack instead of hiding it behind a flat per-minute rate. Buyers pick the LLM, voice, and telephony independently and see each layer’s contribution to the per-minute price — turning model selection into a direct cost lever. Calls are metered to the second with no per-call rounding, though billing continues through silence and hold because the STT engine stays active.

Pricing by product

Retell AI (plans)

Voice agents — component pricing (Default mode)

Voice agents — Speech-to-Speech mode

Chat agents (per AI message)

Add-ons and monthly subscription items

Sales motions across products: self-serve PLG for pay-as-you-go (instant signup, $10 credits, no contract needed to start building), and sales-led for Enterprise (volume pricing, compliance, dedicated support).

Hidden costs : What Retell AI users actually pay

The headline $0.07/min is the floor for the cheapest LLM with no extras. Real bills are shaped by which LLM you pick (a 4x swing from GPT 5.1 to GPT 5.5), the voice (ElevenLabs is $0.04/min vs $0.015/min for platform voices), telephony, per-minute add-ons, and monthly subscription items.

Other things to budget for: billing runs during silence and hold because the STT engine stays active; branded outbound calls add $0.10 per call; PII removal is +$0.01/min and AI Quality Assurance is $0.10/min after the first 100 free minutes; extra concurrency above the 20 free is $8/call/month; and Retell phone numbers ($2/mo) and verified numbers ($10/mo) are recurring.

Want to estimate your own Retell AI bill? Use the Retell AI pricing calculator to model your costs based on usage patterns.

Pricing evolution : Retell AI pricing history and changes

Cadence

Tracked range: 2024–present. Retell has kept a stable true-pay-as-you-go structure since launch; the menu of LLMs, voices, add-ons, and monthly items has broadened while the unbundled per-minute model stayed constant. (Wayback archive access was unavailable for this capture, so historical snapshot dates are based on company milestones rather than archived pricing pages.)

Notable changes

• 2024-02 — Launches (YC W24) with true pay-as-you-go per-minute voice agents and no platform fee. Reaches ~$3M annualized revenue within months.
• 2024-09 — Announces a $4.6M seed led by Alt Capital; pricing stays transparent and self-serve as BPO/enterprise adoption grows.
• 2025–2026 — Expands the LLM and voice menu (GPT-5 family, Claude 4.x, Gemini), adds chat agents (per AI message), an add-on menu (knowledge base, denoising, PII removal, QA, branded calls), and monthly subscription items. Scales to ~$60M ARR by April 2026.

What’s unique : Retell AI’s distinctive pricing mechanics

1. Fully unbundled per-minute stack. Instead of a single opaque per-minute rate, Retell itemizes every voice minute into Voice Infra, LLM, TTS, and telephony. Buyers choose each layer independently and see its contribution, making model and voice selection a direct, transparent cost lever — unusual in a category that usually quotes one blended number.

2. True pay-as-you-go with no contract. No platform fee, no seat minimum, no annual commitment — a deliberate contrast to voice-AI rivals that require annual contracts “before you write a single line of code.” $10 in free credits and 20 free concurrent calls lower the barrier to launch.

3. Second-level metering with honest edge cases. Calls are tracked to the nearest second with no per-call rounding, and Retell publishes exactly when billing applies (during silence/hold, transfers, voicemail, failed calls). This granular, documented metering builds trust around a notoriously fuzzy unit.

Strengths & weaknesses

Billing UX : Retell AI billing controls and transparency

• Billing controls — Fully self-serve: sign up, get $10 in credits, and pay as you scale with no card required to start building. Concurrency and add-ons can be added or removed anytime. Enterprise is invoiced under a custom MSA/DPA.
• Usage visibility — The “Estimate Your Cost” calculator on the pricing page takes monthly call minutes and a chosen LLM, Text To Speech voice, telephony route, and Call Agent Add-ons, then breaks the result into named lines — Cost Per Minute, LLM Cost, Retell Voice Infra, TTS Cost, Telephony Cost, Add-ons, and Total per month (default configuration: $0.115/min, $11.5 for 100 minutes). In-product call analytics and transcripts cover after-the-fact usage, and invoices are available in the Billing tab each cycle.
• Component and mode toggles — The “Detailed Component Pricing” block exposes a Default / Speech-to-Speech toggle for call agents and a per-country telephony picker (~19 Twilio/Telnyx routes), so a buyer can price the exact stack before committing. Concurrency is a self-serve line item — 20 active calls free, $8.00/concurrency/month beyond that, addable or removable anytime.
• Payment options — Card-based self-serve credits for pay-as-you-go; Enterprise is invoiced under custom terms. Telephony can be brought via your own SIP trunk to avoid Retell’s pass-through telephony fee.

Strategic wins : Why Retell AI’s pricing decisions worked

1. Transparency as a wedge against contract-first rivals

By leading with “$0, pay only for what you use” against competitors that demand annual contracts up front, Retell turned pricing into a top-of-funnel advantage. The $10 credits and self-serve launch path converted developers fast — a contributor to ~650% YoY growth. See how AI companies structure pricing.

2. Unbundling the stack to align price with value

Itemizing Voice Infra, LLM, TTS, and telephony lets buyers right-size each layer (cheap LLM for simple flows, premium voice where it matters). This makes the per-minute number feel fair and gives Retell a clean margin on its own $0.055/min infra while passing through model costs. Related: outcome-based pricing trends.

3. Per-minute as the value metric buyers already count

Contact centers already think in agent-minutes and cost-per-minute, so pricing per voice-minute maps directly onto the metric Retell displaces — offshore human agents at $0.30–$0.80/min. Undercutting that by 70–95% makes the ROI obvious. See choosing the right usage metric.

Areas to improve : Gaps in Retell AI’s pricing approach

1. Composite bills are hard to forecast

The flip side of unbundling is that a single voice minute draws from four or five separately priced layers plus add-ons, so estimating a monthly bill requires modeling each one. The calculator helps, but a few representative “recipe” bundles with all-in per-minute prices would reduce buyer effort. See bill shock and cost unpredictability.

2. Billing during silence and hold

Charging for silence and hold time (because STT stays active) is honest and documented, but it can surprise buyers benchmarking against per-spoken-word pricing. Clearer up-front framing of effective cost-per-handled-call would set expectations better.

3. No committed-use discount for self-serve

Self-serve buyers get no volume break — only Enterprise unlocks volume pricing, and that path is sales-led and opaque. A published tiered or committed-use discount would give high-volume PLG accounts a reason to consolidate without a sales call.

Key takeaways

1. Unbundling can be a trust feature. Showing exactly what each layer of a voice minute costs turns pricing transparency into a differentiator in a category that usually quotes one blended rate.
2. True pay-as-you-go beats contract-first in PLG. Removing contracts, fees, and minimums (plus $10 credits) lowered the barrier enough to drive ~650% YoY growth.
3. Price on the metric your buyer already counts. Per-voice-minute maps directly onto the agent-minute economics of the call centers Retell displaces.
4. Document the fuzzy edges. Publishing exactly when billing applies (silence, hold, transfers, voicemail, failed calls) defuses disputes around a notoriously ambiguous unit.
5. Add-ons and monthly items quietly compound. A long menu of small per-minute and per-month charges can erode the simplicity of a “pay only for what you use” headline.

UBP implications

1. Unbundled, pass-through usage pricing aligns vendor margin with buyer choice. Charging a clean margin on your own infra layer while passing through model/voice/telephony costs keeps incentives honest and lets buyers tune spend.
2. Pick a value metric the buyer already budgets in. Voice-minutes (vs. tokens) let a contact-center buyer compare directly against human-agent cost-per-minute — the comparison that closes the deal.
3. Transparency requires a forecasting aid. When the unit is composite, an interactive calculator (or recipe bundles) is what makes usage-based pricing usable. See usage-based pricing strategy.

Sources

• Retell AI pricing page — live capture (accessed 2026-07-22)
• Retell AI seed announcement (accessed 2026-06-09)
• Y Combinator — Retell AI (W24) (accessed 2026-06-09)
• Sacra — Retell AI at $60M/yr, up 650% YoY (accessed 2026-06-09)
• Y Combinator post — $4.6M seed, $3M ARR since February launch (accessed 2026-06-09)

Bottom line

Retell AI is a YC W24 conversational voice-agent platform that turned radical pricing transparency into a growth engine — true pay-as-you-go voice agents at $0.07–$0.31/min, unbundled into Retell Voice Infra ($0.055/min) plus the LLM, voice, and telephony you choose, with chat agents from $0.002/AI msg, $10 in free credits, and 20 free concurrent calls. No contracts, no platform fee, second-level metering with documented billing rules; Enterprise is custom-quoted with volume pricing, dedicated infrastructure, and compliance. That model helped Retell reach ~$60M ARR by April 2026, up roughly 650% YoY. Browse the pricing blueprint for more fully-researched company profiles.

Want to compare Retell AI against other voice-AI companies like Vapi, Bland, or ElevenLabs? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Model menu refresh — Gemini 3.5 Flash added

Jul 2026

Routine addition to the selectable voice LLM menu: Gemini 3.5 Flash at $0.081/min (voice). No change to the unbundled per-minute structure, headline $0.07–$0.31/min voice rate, $0.055/min Retell Voice Infra, $0.002+/msg chat, telephony ($0.015/min), add-ons, or monthly items ($2 phone / $20 SMS / $8 concurrency / $8 KB / $10 verified number). All other LLM, TTS, and chat prices unchanged.

Model menu refresh — Claude 4.6, Gemini 3.0 Flash, GPT 5.x reshuffle

Jun 2026

Routine refresh of the selectable LLM/voice menu (Claude 4.6 Sonnet $0.08/min, Gemini 3.0 Flash $0.027/min added; GPT 5.x 'Recommended' tiers reshuffled) with no change to the unbundled per-minute structure, headline $0.07–$0.31/min voice rate, $0.055/min Retell Voice Infra, $0.002+/msg chat, telephony ($0.015/min), add-ons, or monthly items ($2 phone / $20 SMS / $8 concurrency / $8 KB / $10 verified number).

Unbundled per-minute stack + chat agents + add-ons

Jun 2026

Current structure: voice agents at $0.07–$0.31/min ($0.055/min Retell Voice Infra + LLM + TTS + telephony), chat agents from $0.002/AI msg, $10 free credits and 20 free concurrent calls, an expanded add-on menu (knowledge base, denoising, PII removal, QA, branded calls), monthly subscription items, and a custom-quoted Enterprise plan with volume pricing, dedicated servers, HIPAA/BAA and SSO.

$4.6M seed announced

Sep 2024

Announced a $4.6M seed led by Alt Capital (with Y Combinator, Carya Venture, and 20+ founders/operators) while sustaining the unbundled pay-as-you-go pricing. Pricing remained transparent and self-serve as the company scaled into BPO and enterprise contact-center accounts.

Launch — pay-as-you-go voice agents

Feb 2024

Retell AI (YC W24) launches in February 2024 with a true pay-as-you-go model: per-minute AI voice agents composed from Retell's voice infrastructure plus a chosen LLM, TTS, and telephony — no contracts, no platform fee. Reached ~$3M annualized revenue within months.