ElevenAgents Puts the Base Cost of Voice AI at $0.08 per Minute
ElevenLabs has broken down the economics of ElevenAgents: on paid plans, the platform itself costs $0.08 per minute, but that is not the final price of a conversation. The language model and phone connectivity are billed separately, and the plan must also be selected based on the number of concurrent calls. For businesses, the key takeaway is simple: calculate the cost per resolved request, including integrations, quality control, and the handoff of complex cases to humans.
AI-processed from ElevenLabs Blog; edited by Hamidun News
Voice AI is often sold with a single attractive per-minute figure. The problem is that this figure rarely reflects the entire bill. ElevenLabs has broken down the cost of its ElevenAgents service and, in effect, proposed a useful formula for any similar project: platform plus language model plus telephony. Then come the expenses that never appear in the marketing calculator.
Across all paid ElevenAgents plans, the platform's base price is $0.08 per minute. This includes speech recognition and synthesis, dialogue management, a knowledge base, and search across it. However, LLM usage and connection to the telephone network are billed separately. Comparing services solely by plan price is therefore much like choosing a cloud server based on the cost of its chassis while ignoring the processor, traffic, and data storage.
What the Actual Bill Includes
The formula works as follows: multiply the number of minutes by the platform rate, then add LLM and telephony costs. If a company handles 8,000 five-minute calls, that amounts to 40,000 minutes per month. The ElevenAgents platform alone will cost $3,200. Bills for the model, phone numbers, and outbound or inbound traffic come on top of that.
The choice of LLM also affects the price. Checking an order status or rescheduling an appointment usually does not require the most expensive reasoning model. It may be more useful for technical diagnostics or a complex sale. A practical approach is to split the workflows: send simple requests to a low-cost model and route complex ones to a more capable model or a human.
Telephony also changes the economics. A voice widget embedded in a website or app does not require a connection to the conventional telephone network and eliminates carrier costs. For traditional calls, the rate depends on the country, direction, number type, and telecommunications provider.
Why Minutes Are Not the Only Metric
ElevenAgents plans impose limits on concurrent conversations. For example, the Pro plan supports up to 20 calls, Scale up to 30, and Business up to 40. This means choosing a plan based solely on total monthly volume is not enough. If 50 customers call simultaneously in the morning, a large remaining balance of minutes will not prevent a queue or an additional peak-capacity charge. ElevenLabs lists a rate of $0.16 per minute for calls above standard capacity.
A product team therefore needs at least four metrics: average conversation duration, peak concurrent calls, the share of requests resolved without a human, and the cost of one fully resolved request. The last metric is more useful than an advertised per-minute price: a long, inexpensive conversation followed by an escalation may cost more than a short dialogue using a more capable model.
In its own example, ElevenLabs uses 40,000 minutes and an assumed human labor cost of $40 per hour. If the agent can independently resolve 60% of requests, the savings before LLM and telephony costs amount to approximately $12,800 per month. This is neither a forecast nor a universal benchmark: the result depends entirely on wages, call patterns, and the actual rate of successful resolutions.
What to Include in a Pilot
The marketing estimate does not account for CRM integration, knowledge-base configuration, testing, conversation analysis, workflow updates, security, or the work of employees handling complex cases. These are precisely the items that can turn an impressive prototype into an expensive production project.
Before launch, it is worth selecting one high-volume, easily measurable workflow—rescheduling an appointment, for example—and running a pilot on real traffic. Platform, LLM, and connectivity costs should be tracked separately, along with the number of escalations, errors, and repeat calls. The comparison should then focus not on minutes, but on the cost of a successfully completed task.
The conclusion for developers and founders is straightforward: the $0.08-per-minute rate is the bottom layer of the estimate, not the price of a finished voice employee. The solution makes sense when the agent reliably completes routine operations, handles peak demand, and leaves humans only those conversations that genuinely require experience, accountability, or empathy.
- Nov 12, 2026Claude in Chrome requires approval for every click on category 3 sites
- Nov 12, 2026Norwest raises record $4 billion fund, betting on agents, voice and physical AI
- Nov 12, 2026AWS shows how 100% retrieval precision can hide an AI agent's failure
- Nov 12, 2026Garage Eight dropped peer approvals for 69 MRs in an internal service using AI review
Key Questions Answered
What the Actual Bill Includes?
The formula works as follows: multiply the number of minutes by the platform rate, then add LLM and telephony costs. If a company handles 8,000 five-minute calls, that amounts to 40,000 minutes per month. The ElevenAgents platform alone will cost $3,200. Bills for the model, phone numbers, and outbound or inbound traffic come on top of that. The choice of LLM also affects the price. Checking an order status or rescheduling an appointment usually does not require the most expensive reasoning model.
Why Minutes Are Not the Only Metric?
ElevenAgents plans impose limits on concurrent conversations. For example, the Pro plan supports up to 20 calls, Scale up to 30, and Business up to 40. This means choosing a plan based solely on total monthly volume is not enough. If 50 customers call simultaneously in the morning, a large remaining balance of minutes will not prevent a queue or an additional peak-capacity charge. ElevenLabs lists a rate of $0.16 per minute for calls above standard capacity.
What to Include in a Pilot?
The marketing estimate does not account for CRM integration, knowledge-base configuration, testing, conversation analysis, workflow updates, security, or the work of employees handling complex cases. These are precisely the items that can turn an impressive prototype into an expensive production project. Before launch, it is worth selecting one high-volume, easily measurable workflow—rescheduling an appointment, for example—and running a pilot on real traffic.
Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews
Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.