ElevenLabs Agents Pricing and Voice Agent Cost per Minute

How much does ElevenLabs Agents cost?

ElevenLabs Agents costs $0 on the free plan, then $6 a month on Starter, $22 on Creator ($11 for the first month), $99 on Pro, $299 on Scale and $990 on Business. Enterprise is custom pricing. Those plans include 15, 75, 275, 1,238, 3,738 and 12,375 minutes of calls a month, and 4, 6, 10, 20, 30 and 40 concurrent calls. Scale adds 3 workspace seats, and Business adds 10 seats and 10 professional voice clones.

Every plan bills the same $0.08 for each minute past the included allowance and $0.003 for each text message an agent sends. Prices exclude taxes. ElevenLabs bills these plans by call minute, not from the shared credit pool its creative tools use, which is why a minute on an agent costs the same whichever voice you pick.

The page also lists a pay as you go option next to the free plan. ElevenLabs introduced pay as you go across its API and agents products in May 2026, alongside the cut from $0.10 to $0.08 a minute on Starter. Some third party pages still quote the old $0.10 rate, so budget from the vendor's own page. ElevenLabs also runs a Startup Grants Program, which gives eligible startups 12 months free with 33 million characters to build and test an agent.

How much does an ElevenLabs voice agent cost per minute?

An ElevenLabs voice agent costs $0.08 a minute for the platform itself, plus the language model, plus the phone line. The platform part covers hosting, speech to text and text to speech. It is the same on every plan once you are past the included minutes, and the included minutes work out to roughly the same rate: $99 for 1,238 minutes on Pro is $0.08 a minute, and $990 for 12,375 on Business is $0.08 too.

The model is the variable. ElevenLabs lets you choose from dozens of hosted models from OpenAI, Anthropic, Google and its own catalog, or connect your own model through a server integration. Its docs say it "passes through third-party LLM costs at the provider's published rate, with no markup", and that certain Gemini and Claude models are priced at Vertex AI regional rates, which are 10 percent higher for input, output and cache tokens, with US and EU traffic billed at those rates. A short sales call with a small, fast model adds a fraction of a cent a minute. A long call on a frontier model can add more than the $0.08 platform rate.

The phone line is the third part. ElevenLabs connects to Twilio or a SIP trunk and adds no telephony fee of its own, so the carrier bills you directly. On Twilio's US voice price list, read on October 22, 2026, an outbound call costs $0.014 a minute, a local number $1.15 a month, answering machine detection $0.0075 a call and recording $0.0025 a minute. A realistic all in minute for an outbound sales agent is therefore about $0.09 to $0.10 before the model, and higher with a large one.

What is ElevenLabs burst pricing?

Burst pricing lets your agents run above the concurrency limit on your plan during a spike, up to three times that limit, with every excess minute billed at double the standard rate. On ElevenAgents that is $0.16 a minute instead of $0.08. It is a setting you enable, not something that happens by default.

For outbound calling this matters more than it does for inbound support. A dialing campaign is a burst by design: you load a list and want as many calls running at once as the plan allows. A Pro plan with 20 concurrent calls can run up to 60 with burst enabled, but calls 21 to 60 cost twice as much. If your campaign regularly needs 30 lines, moving to Scale at $299 with 30 concurrent calls is cheaper than running Pro in burst for more than a few hundred minutes a month.

The quick rule from our arithmetic: compare the burst minutes you expect against the price gap to the next plan. Pro to Scale is $200 a month, which buys 2,500 minutes of the extra $0.08 burst premium. If you burst more than that, upgrade.

What 5,000 and 20,000 outbound minutes cost on ElevenLabs

These worked months are our arithmetic on the October 22, 2026 rates, for outbound calling in the US through Twilio, before the language model.

At 5,000 minutes a month, Pro costs $99 plus 3,762 extra minutes at $0.08, which is $399.96. Scale costs $299 plus 1,262 extra minutes, which is also $399.96. Creator lands at $400 too. Because the included minutes and the overage are priced the same, the platform bill for 5,000 minutes is about $400 on any plan. Twilio adds about $70 for the outbound minutes and a few dollars for numbers, so the month is roughly $475 before the model. What you are choosing between plans is concurrency: 10 lines on Creator, 20 on Pro, 30 on Scale.

At 20,000 minutes a month, Business costs $990 plus 7,625 extra minutes, which is $1,600, and Scale lands at the same $1,600. Twilio adds about $280. The month is roughly $1,900 before the model, or about $0.095 a minute.

The cost that is not on either line is the build. Somebody has to write the agent's prompt and call flow, connect the CRM and calendar, load and scrub the lists, set calling windows by time zone, handle voicemail and answering machines, log outcomes and keep the agent current as your offer changes. For a team with an engineer who wants that job, ElevenLabs is one of the cheapest ways to run a voice agent. For a sales team without one, the build is the real price.

What the ElevenLabs Agents price does not include for outbound sales

ElevenLabs sells a voice agent platform, and it is a good one. It does not sell an outbound sales program, and several things a US outbound team needs are your job on top of the minute rate.

Compliance logic is the biggest. Calling consumers with an AI voice is treated as an artificial or prerecorded voice call under the TCPA, which needs prior express consent, and calls to business lines still have to respect DNC requests, calling hours and state rules. On a voice platform, you build the consent check, the federal and state DNC scrub, the time zone windows and the AI disclosure into your own stack. We cover the rules in detail on is AI cold calling legal.

Caller ID reputation is the second. A new number that places hundreds of AI calls a day gets labeled as spam within days on the major carrier analytics systems, and the platform does not manage that for you. Number rotation, registration and branded caller ID sit with you and your carrier, and we explain the options on branded caller ID.

The third is the sales layer: qualification criteria, objection handling written for your market, meeting booking onto the right rep's calendar and CRM write back. ElevenLabs gives you a workflow builder, a knowledge base and tool calls to build those pieces, which is real flexibility, and it is still work your team owns.

ElevenLabs vs Vapi vs Retell pricing for a voice agent

The three platforms price the same minute in different ways, so the cheapest one depends on how much you want bundled.

ElevenLabs charges $0.08 a minute with its own speech models inside the rate, then passes the language model through at cost and leaves the carrier to you. Vapi charges $0.05 a minute for hosting and leaves speech, model and telephony to whichever providers you plug in, which can come out cheaper or dearer depending on your stack; we break it down on Vapi pricing. Retell publishes a range of $0.07 to $0.31 a minute as a stack, and the width of that range is almost entirely the model you choose. Bland is the only one of the four that includes the model, transcription and voice in a single rate, from $0.14 a minute.

On our reading, ElevenLabs is the best value of the four when voice quality is the priority, because its speech models are the product and they are inside the $0.08. Vapi suits an engineering team that wants to swap every component. Retell suits a team that wants predictable stack pricing and generous free concurrency. Our full side by side is on AI voice agent pricing, and Vapi vs Retell AI goes deeper on those two.

ElevenLabs Agents vs a done for you AI SDR

On price per minute, ElevenLabs is far cheaper than we are, and we would rather say so plainly. Our Starter plan is $499 a month for one AI SDR seat with 250 connected minutes, or $249.50 a month on yearly billing. That is several dollars per connected minute against roughly ten cents on ElevenLabs.

What the difference buys is the finished outbound motion. Our AI SDR is already built for cold and warm outbound: it dials, says it is an AI at the start, holds the conversation, handles the common objections, qualifies on your criteria, books the meeting onto your calendar and writes the outcome back to the CRM. DNC suppression and calling windows are checked at dial time. Nobody on your team writes a prompt, wires a carrier or maintains a call flow, and the price is one published monthly figure with no model or carrier bill on the side.

So the honest split is this. If you have an engineer, a voice use case of your own and time to build, ElevenLabs Agents is one of the best platforms to build on. If what you want is meetings on the calendar from outbound calls this month, without building anything, compare our plans on the pricing page or see how the agent works on AI voice agent for sales.