ElevenLabs Agents pricing explained: tiers, the $0.08 minute, and what's billed on top

ElevenLabs Agents — the conversational AI product from the voice-technology company ElevenLabs — prices differently from most voice-agent platforms: you buy a subscription tier bundling a monthly allotment of "voice agent" call minutes with a concurrency limit, then pay a flat overage beyond the allotment. The published numbers are unusually consistent once you line them up. But two components sit outside them entirely: per the ElevenLabs Agents pricing page (accessed 2026-08-19), "The LLM model and any telephony are billed separately on top, based on usage." This guide walks through what the public pages let you compute — and where the arithmetic has to stop.

How ElevenLabs Agents pricing works

The pricing page lists six self-serve tiers plus Enterprise. Each tier bundles included call minutes and a concurrent-call limit:

Source: elevenlabs.io/pricing/agents, accessed 2026-08-19. The Business tier also lists 10 Professional Voice Clones; Enterprise adds custom SSO, "BAAs for HIPAA customers," and "Custom terms & assurance around DPA/SLAs."

Beyond the allotment, three usage rates apply, identical on every self-serve tier:

• Overage: $0.08 per additional call minute — the page confirms no tier-specific variation.
• Burst: $0.16 per minute for calls above your tier's concurrency limit.
• Text messages: $0.003 each.

The pattern worth noticing (arithmetic ours, illustrative): the included allotments price out at almost exactly $0.08 per minute — 1,238 × $0.08 = $99.04 on Pro, 3,738 × $0.08 = $299.04 on Scale, 12,375 × $0.08 = $990.00 on Business. The effective ElevenLabs-layer rate is about $0.08 per connected minute on every paid self-serve tier (the Free tier's 15 included minutes are simply $0); moving up a tier mainly buys concurrency and features, not a cheaper minute.

What's included vs. billed separately

The tier minutes and overage cover the ElevenLabs platform layer — the agent runtime and ElevenLabs' own voices. Two components are explicitly outside it:

Sources: elevenlabs.io/pricing/agents, accessed 2026-08-19. You can also pick from supported LLMs or "bring your own custom model" per the platform overview, accessed 2026-08-19 — either way, the model cost is a pass-through, not part of the minute rate.

Worked example: 10,000 minutes — and where the public arithmetic stops

Take a common small-business workload: 2,000 calls × 5 connected minutes = 10,000 connected minutes per month, peaking under 10 simultaneous calls. All arithmetic is illustrative.

That volume exceeds every included allotment below Business, so the platform layer computes four ways:

Creator: $22 + (10,000 − 275) × $0.08 = $22 + $778 = $800.00/month Pro: $99 + (10,000 − 1,238) × $0.08 = $99 + $700.96 = $799.96/month Scale: $299 + (10,000 − 3,738) × $0.08 = $299 + $500.96 = $799.96/month Business: $990/month flat — 12,375 included minutes cover the volume with no overage.

Because included minutes price out at the overage rate, Creator, Pro, and Scale all converge on the same ~$800 total. One concurrency note: Creator's limit is exactly 10 concurrent calls — this workload's under-10 peak fits, and burst billing ($0.16/minute) applies only to minutes above the limit — but a workload that grows to touch 10+ simultaneous calls starts paying burst rates, which is what stepping up to Pro (20 concurrent) buys headroom against. The cheapest published route is therefore about $800/month for the ElevenLabs platform layer.

And that is where the public arithmetic stops. The complete monthly total cannot be computed from ElevenLabs' public pages, because the remaining components carry no published fixed rates: the LLM bills based on usage against your credits and varies by model, and telephony through external providers bills "at cost." That is simply what the public pages support. A buyer pricing this workload knows the platform layer (~$800–$990) and must then add:

1. LLM usage across 2,000 five-minute conversations, at rates depending on the model selected (or their own provider's bill with a custom model), plus
2. Telephony — carrier minutes for 10,000 minutes through their external provider.

The safe statement from public numbers alone: ~$800 is the floor, not the price.

Gotcha checklist

• Tier choice is about concurrency, not the minute rate. Every paid self-serve tier works out to ~$0.08/minute at the platform layer. Model your busiest hour, not your monthly average — the concurrency limit (4 to 40 by tier) is what you're actually buying, and minutes above it bill at double the rate ($0.16).
• The LLM is not in the minute. Model fees are deducted from your ElevenLabs credits based on usage and vary by model. Price your intended model at your conversation length before forecasting.
• Telephony is separate, with no published rate. External-provider telephony is listed "at cost" — budget your carrier's per-minute price on top for every connected minute.
• HIPAA lives on the Enterprise tier. BAAs "for HIPAA customers" appear under Enterprise (custom pricing), so a HIPAA requirement moves you off published pricing. And per the batch calling docs (accessed 2026-08-19), Zero Retention Mode "cannot be enabled for batch calls."
• Right-size in both directions. The subscription is fixed: run 4,000 minutes on Business and you've paid $990 for minutes Scale would have covered for about $320. Match the tier to realistic volume, not aspirational volume.

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 from public numbers: $200, $500, or $900 per month by engine (illustrative arithmetic; premium voice or language paths can add up to 3¢/minute, hold bills a flat 2¢/minute, and telephony depends on setup). The structural contrast: ElevenLabs' ~$0.08 platform minute excludes the model and telephony, while ThunderPhone's engine rate includes the model and voice in the printed number. ThunderPhone is GDPR and HIPAA compliant; customers remain responsible for configuring their workflows, data handling, recordings, and notices for their own obligations.

FAQ

How much does ElevenLabs Agents cost per minute?

About $0.08/minute at the platform layer on every paid self-serve tier — before the LLM and telephony. Included tier minutes price out at ~$0.08 and additional minutes bill exactly $0.08, but the language model and telephony are billed separately on top based on usage, so the complete per-minute cost is not published as a single number. ElevenLabs Agents pricing, accessed 2026-08-19.

What's the difference between the ElevenLabs tiers?

Mostly concurrency and features, not the minute rate. Tiers run from Free ($0, 15 minutes, 4 concurrent calls) to Business ($990, 12,375 minutes, 40 concurrent), and each paid allotment works out to ~$0.08/minute; Enterprise adds custom SSO and HIPAA BAAs. ElevenLabs Agents pricing, accessed 2026-08-19.

Does ElevenLabs Agents pricing include the LLM and phone calls?

No. The pricing page states "The LLM model and any telephony are billed separately on top, based on usage" — model fees are deducted from your ElevenLabs credits and vary by model, and external-provider telephony is listed "at cost" with no published per-minute price. ElevenLabs Agents pricing, accessed 2026-08-19.

What happens if I exceed my concurrency limit?

Minutes above your tier's concurrent-call limit bill at the $0.16 burst rate — double the standard overage. Concurrency runs from 4 calls on Free to 40 on Business, so peak simultaneous volume, not monthly totals, should drive tier selection. ElevenLabs Agents pricing, accessed 2026-08-19.

What would 10,000 minutes a month cost on ElevenLabs Agents?

About $800/month at the platform layer (Creator, Pro, or Scale plus overage; $990 flat on Business) — plus LLM and telephony costs the public pages don't let you compute. For contrast, ThunderPhone's all-in rates put the same workload at $200/$500/$900 by engine, illustratively, with no subscription — the model and voice are inside the engine rate, while telephony depends on setup.

Sources and freshness

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