ElevenLabs agent pricing: real cost per minute and monthly estimates

This page may contain affiliate links from which Blur Studio earns commissions.

ElevenLabs Agents, ElevenLabs' platform for building conversational AI voice agents that can answer and make phone calls, prices its usage on included call minutes per plan, plus a per-minute overage rate once you exceed them. This guide breaks down the current vendor pricing, works through real monthly examples at a few call volumes, and compares the result to what a human answering service typically costs.

Vendor pricing

Plan	Monthly price	Included minutes	Concurrent calls
Free	$0	15 minutes	4
Starter	$6	75 minutes	6
Creator	$22	275 minutes	10
Pro	$99	1,238 minutes	20
Scale	$299	3,738 minutes	30
Business	$990	12,375 minutes	40
Enterprise	Custom	Custom	Custom

Beyond included minutes, standard calls bill at $0.08 a minute, rising to $0.16 a minute if you exceed your plan's concurrent call limit. Text-based agent messages, separate from voice minutes, bill at $0.003 each. Two costs sit entirely outside this structure: LLM usage is billed separately based on the model you choose, and telephony (the actual phone connection through a provider like Twilio) passes through at that provider's own rate with no ElevenLabs markup.

Example calls per month

Here's what a few realistic monthly volumes work out to, using just the ElevenLabs agent-minute cost above (LLM and telephony costs are additional and vary by provider and model, so they're not included in these figures):

• A small business, roughly 200 calls a month averaging 3 minutes each (600 minutes): on Creator ($22, 275 included minutes), the 325 extra minutes cost $26 at the overage rate, for a total of about $48 a month.
• A growing business, roughly 1,000 calls a month averaging 4 minutes each (4,000 minutes): on Pro ($99, 1,238 included minutes), the 2,762 extra minutes cost about $221, for a total of roughly $320. Interestingly, Scale ($299, 3,738 included minutes) works out to almost exactly the same total, about $320, since the 262 extra minutes only add about $21. This isn't a coincidence: ElevenLabs' included-minute jumps between tiers are priced consistently at the same $0.08-a-minute marginal rate, so above a certain volume, the plan you pick is really a decision about concurrent call capacity, not raw minute cost.
• A larger team running roughly 10,000 minutes a month: Business ($990, 12,375 included minutes) covers this with no overage at all. Scale plus overage, by contrast, comes out cheaper on paper, about $299 plus 6,262 extra minutes at $0.08 ($501), for a total around $800, so it's worth checking both the flat higher tier and a lower tier plus overage before assuming the bigger plan is the better deal, as long as Scale's 30 concurrent call limit is actually enough for your volume.

In every case, add your LLM provider's cost and your telephony provider's per-minute rate on top of these figures for a true total. My separate guide to building an ElevenLabs phone agent covers how those two additional costs actually get incurred.

Comparison with human answering services

A traditional human answering service or virtual receptionist typically bills per minute or per call, and its rate generally runs well above the $0.08-a-minute overage figure above once you include the human labor cost built into that price, often several times higher on a straight per-minute basis. The honest trade-off isn't just price, though. A human receptionist can genuinely handle a confused or upset caller, an unusual request, or a judgment call that an AI agent's knowledge base simply wasn't built to cover, something I flag as the real risk point in AI phone agents in my agent setup guide. For high-volume, repetitive call types (booking confirmations, hours and location questions, simple intake), an AI agent is usually the better economics. For calls where getting it wrong has real consequences, a blended setup, AI handling the routine volume with a clean handoff to a human for anything it can't confidently answer, is often the more sensible design than an all-or-nothing choice. AI receptionist vs human receptionist compares cost and quality in more detail.

Estimating the LLM and telephony costs you need to add

The figures above cover only ElevenLabs' own agent-minute charge. In practice, two more costs sit on top of every number above. LLM cost depends entirely on which model you connect the agent to, a lighter, faster model costs less per conversation turn than a larger, more capable one, and since an agent generates many turns per call, this cost scales with conversation length and complexity, not just call count. Telephony cost depends on your provider, Twilio or otherwise, and is typically a small per-minute fee on top of the ElevenLabs agent charge, passed through at the provider's own rate with no ElevenLabs markup. For budgeting purposes, it's more useful to run a short pilot, a few hundred real minutes, measure the actual combined bill across all three costs, and extrapolate from that real number, than to estimate LLM and telephony costs in the abstract before you've seen a single real conversation's actual billing breakdown. Building an AI receptionist: tools, steps and cost lists the full stack.

When the cheaper plan isn't actually cheaper

The worked examples above show that total cost at a given volume often lands close to identical across adjacent tiers, which means the decision that actually matters is concurrency, not the sticker price of the plan. If your call volume is heavily concentrated in a few peak hours (a service business getting most calls during business hours, for instance) rather than spread evenly across the day, you need enough concurrent call capacity to cover your actual peak, not just your monthly total divided evenly. Underestimating concurrency shows up as callers hitting a busy signal or a fallback message during your busiest hours, even though your total monthly minutes looked comfortably within a lower tier's included allowance. Size the plan to your peak-hour call pattern, not just your monthly average.

A simple way to track spend after launch

Once an agent is live, the single most useful habit is checking actual usage against your plan's included minutes on a fixed schedule, weekly for the first month, then monthly once the pattern stabilizes, rather than waiting for a surprising invoice to prompt a review. Watch for a rising trend in average call length specifically, not just call count, since a few calls running long due to a confused caller or a knowledge base gap can quietly push your overage cost up even if your total call count hasn't changed. Catching that early, and fixing the underlying knowledge base gap causing the longer calls, usually costs less effort than discovering it three months later in a bill that's crept up without an obvious single cause.

How I'd size this for a client

When I set up an agent for a client, I estimate real call volume and average call length from their existing phone data first, not a guess, then pick the plan tier based on concurrency needs rather than defaulting to whichever tier happens to include enough minutes on paper. Get in touch if you want this sized properly against your actual call volume before you commit to a plan.