ElevenLabs Agents Pricing (2026): Minutes, LLM Costs, Phone Numbers and Real Monthly Bills

ElevenLabs Agents plans, minutes and concurrency#

• Extra minutes: $0.08 each on every paid plan, per ElevenLabs' pricing FAQ.
• Burst: if you switch it on, an agent can take up to three times your concurrency limit; calls above the limit cost $0.16 a minute, double the standard rate.
• Text messages: $0.003 each.
• Annual billing: 10 months for 12 across ElevenLabs plans, which lowers the effective rate to about $0.067 a minute on included minutes.

What a month actually costs: three examples#

Because the minute price is flat, budgeting is mostly about volume and peak concurrency. These are our calculations on the published rates, before the LLM and your phone provider.

The agency row shows the pattern: at 3,000 minutes, Creator and Pro cost the same, so the only question is whether 10 simultaneous calls is enough. If your calls bunch up, for example everyone phones a restaurant at 6pm, buy for the peak, or enable burst and accept $0.16 a minute on the overflow.

The LLM is billed separately#

ElevenLabs runs the voice, speech recognition and turn-taking; the "thinking" comes from a language model you choose. ElevenLabs says it passes third-party LLM costs through "at the provider's published rate, with no markup", with one exception: some Gemini and Claude models priced at Vertex AI regional rates cost 10% more on input, output and cached tokens. You can pick OpenAI GPT models, Google Gemini, Anthropic Claude, open models hosted by ElevenLabs (such as Qwen, GLM and DeepSeek variants), or your own endpoint.

ElevenLabs does not publish a per-minute LLM figure, because it depends on the model and how much the agent talks. As a reference point, Retell lists a small model at about $0.008 a minute and a large one at about $0.04 (see our voice agent platform comparison). On that range, the LLM adds 10% to 50% to the $0.08 minute. One inconsistency to know about: the pricing FAQ says LLM fees "will be deducted from your ElevenLabs credits", while also saying agent plans are "billed by call minutes, not by the shared credit pool". Check your usage page after a test day before you scale.

Phone numbers: bring your own#

• No ElevenLabs telephony fee. The pricing page says ElevenLabs "does not charge extra for telephony". Your carrier bills you directly.
• Twilio: native integration. You import a number you bought in Twilio into the ElevenLabs dashboard; purchased numbers handle inbound and outbound calls, while a Twilio Verified Caller ID works for outbound only.
• SIP trunk: supported, so you can use another carrier or an existing phone system.
• Batch outbound calls and WhatsApp are listed as channels. ElevenLabs does not publish separate prices for them on the pricing page.
• ElevenLabs' pages do not offer to sell you a phone number directly; plan on a Twilio or SIP account.

If you are calling people rather than answering calls, read the consent rules first: our AI outbound calling guide covers the FCC and TCPA position on AI voices.

Languages, compliance and limits to check first#

• Languages: the agents product page says 90+ languages; the ElevenAgents docs say 31 languages in one place and 70+ in another. Test your exact language and accent on the Free plan before you commit.
• Compliance (as stated by ElevenLabs): SOC 2 Type II, ISO 27001, ISO 42001, PCI DSS Level 1, HIPAA and GDPR, regional data residency (US, EU and India on Enterprise) and a zero-retention mode. A HIPAA BAA is listed under Enterprise only.
• EU data residency makes some LLMs unavailable, per the LLM docs.
• Not stated: whether silence or hold time is billed differently, and whether test calls and simulations count as minutes. Assume every connected minute counts.

Ways to pay less#

• Start on Free: 15 minutes and 4 concurrent calls is enough to build and test one agent.
• First-month offers: Starter at $1 and Creator at $11 for the first month were on the pricing page on November 10, 2026, with no end date given.
• Annual billing: 10 months' price for 12.
• Startup grants: ElevenLabs gives eligible startups the Scale plan free for 12 months (stated as 33 million characters). Eligibility includes a monetised product, under 25 employees, a business email and no agencies or consultancies. Details are in our ElevenLabs discounts guide.
• Keep calls short: with a flat per-minute price, a tighter script and a quick hand-off to a human are the biggest saving.

ElevenLabs Agents vs Retell, Vapi and Bland#

On price per minute the main platforms are close: roughly $0.075 to $0.14 a minute before phone charges. ElevenLabs is the strongest choice when the voice matters (a brand voice, a cloned voice or many languages) and the cheapest place to get ElevenLabs voices: Retell charges $0.04 to $0.10 a minute for them on top of its own $0.055 platform fee. Retell gives more concurrency on its entry tier (20 calls), Vapi gives developers the most control over each component, and Bland bundles the LLM into one higher rate. Full numbers are in our voice agent platform comparison and our Vapi alternatives guide.

If you would rather not build at all, an off-the-shelf AI receptionist may cost less in staff time; our build vs buy guide does that maths. To try ElevenLabs Agents, the Free plan is enough for a first agent.