Voice AI charges are usage-based — you only pay for what you use. Every call can generate charges across up to four cost components. This guide breaks down each charge type clearly so you can estimate costs, plan your budget, and scale with confidence.
1
The Four Cost Components
⚙️ Voice Engine Fee
Core processing charge for running AI during every call. Applied per minute, regardless of voice or model.
Rate: $0.045 / minute
Rate: $0.045 / minute
🔊 Text-to-Speech (TTS) Fee
Cost to generate the AI’s spoken voice via ElevenLabs. Charged per minute based on the model selected.
Rate: $0.035 – $0.170 / min
Rate: $0.035 – $0.170 / min
🧠 AI Model (LLM) Token Fee
Cost for the AI to “think” — reading conversations and generating responses. Charged per 1M tokens processed.
Rate: $0.10 – $2.50 / 1M tokens
Rate: $0.10 – $2.50 / 1M tokens
📞 Phone System Fee
Standard telephony charges that apply to every phone call — separate from AI costs. Varies by call type and destination.
Rate: Varies by carrier / call type
Rate: Varies by carrier / call type
2
Voice Provider Rates (TTS) — ElevenLabs
| Voice Provider | Model | Rate / Min | Best For |
|---|---|---|---|
| ElevenLabs | V2.5 | $0.035 | ✓ Great Value High-quality natural voice |
| ElevenLabs | V3 | $0.170 | ★ Premium Ultra-realistic, human-like voice |
Which ElevenLabs Model Should I Choose?V2.5 delivers excellent natural-sounding voice at a fraction of V3’s cost — ideal for most business use cases. Upgrade to V3 when an ultra-realistic, human-like experience is a priority.
3
AI Model Rates (LLM Tokens)
| Tier | Provider | Model | Price / 1M Tokens |
|---|---|---|---|
| Economy | OpenAI | GPT-4.1 Nano | $0.10 |
| Standard | Anthropic | Claude 3.5 Haiku | $0.80 |
| ★ Premium | OpenAI | GPT-4o | $2.50 |
Token Usage NoteToken costs per call are typically very small. Longer conversations, larger knowledge bases, and more complex prompts use more tokens. Model pricing may change as providers update their rates.
4
How Your Total Cost Is Calculated
⚡ Total Cost Formula
Total Cost = (Minutes × Voice Engine Rate)
+ (Minutes × ElevenLabs TTS Rate)
+ LLM Token Cost
+ Phone System Charges
+ (Minutes × ElevenLabs TTS Rate)
+ LLM Token Cost
+ Phone System Charges
💡 What does your Voice AI budget actually get you? This is what you can expect to spend on your pay-per-use AI communication. Adjust your monthly budget and average call length below to see how many AI-handled calls your investment covers.
Select Tier:
$100
$25$500
Set your estimated monthly Voice AI spend.
5 min
1 min20 min
How long is a typical AI-handled call for your business?
Your budget covers approximately
1,250
AI Voice Minutes
Estimated Calls
250
Cost Per Call
$0.40
Cost Per Minute
~$0.080
Calls / Day (30-day mo.)
8.3
📊 Per-Minute Cost Breakdown — Economy Tier
⚙️ Voice Engine$0.045 / min
🔊 TTS — ElevenLabs V2.5$0.035 / min
🧠 AI Model — GPT-4.1 Nano~$0.000 / min
💵 Total AI Cost~$0.080 / min
* Estimates are based on AI costs only. Phone system charges are billed separately and not included. Actual token usage may vary based on conversation length and knowledge base size.
5
Inbound vs. Outbound — Key Differences
| Feature | Inbound Voice AI | Outbound Voice AI |
|---|---|---|
| Covered by AI Employee Unlimited ($97/mo) | ✅ Yes | ❌ Always pay-per-use |
| Phone System Charges Apply | ✅ Yes | ✅ Yes |
| Voice Engine + TTS Fees | Pay-per-use (if not on Unlimited) | Always pay-per-use |
| LLM Token Fees | Pay-per-use | Pay-per-use |
6
Why Two Calls Can Cost Different Amounts
Even if two calls are the same length, the final cost may vary based on:
🎙️ ElevenLabs Model Selected
V3 costs significantly more per minute than V2.5. Choosing the right model directly impacts your TTS costs.
🧠 AI Model Selected
GPT-4o (Premium) costs 25x more per token than GPT-4.1 Nano (Economy). Match the model to your call complexity.
💬 Conversation Complexity
More back-and-forth exchanges, longer AI responses, and detailed prompts all use more tokens.
📚 Knowledge Base Size
Larger knowledge bases send more context to the AI model as input tokens on every call.
7
Tips to Manage Your Costs
| Strategy | Expected Impact |
|---|---|
| ✅ Start with ElevenLabs V2.5 | Save ~79% on TTS vs. V3 while delivering high-quality voice |
| ✅ Start with GPT-4.1 Nano (Economy tier) | Reduces LLM token costs by up to 96% vs. GPT-4o |
| ✅ Keep AI prompts and instructions concise | Fewer input tokens per call = lower LLM cost |
| ✅ Consider AI Employee Unlimited for Inbound | Flat $97/mo covers Inbound Voice AI at higher call volumes |
| ✅ Monitor usage via the AI Usage Dashboard | Spot high-cost calls early and fine-tune your agent setup |
8
Frequently Asked Questions
Will I be charged if no one answers?
Charges typically begin when a call connects. Unanswered or failed calls generally do not generate Voice AI charges.
Is there a monthly minimum charge?
No. Pay-per-use means you only pay when calls occur. There is no monthly minimum for Voice AI usage.
Which AI model tier should I start with?
Most businesses start with the Economy tier (GPT-4.1 Nano) for straightforward call handling. Upgrade to Standard or Premium only if your calls require more advanced reasoning or nuanced responses.
Where can I see a breakdown of my usage and charges?
The AI Usage Dashboard provides a full breakdown of charges by call, product, and AI model — accessible at the agency level for agency admins.
Does the AI Employee Unlimited plan eliminate all Voice AI charges?
No. AI Employee Unlimited ($97/mo) covers Inbound Voice AI usage, but Phone System charges still apply. Outbound Voice AI, Voice AI Widget, and Voice AI Prompt Optimizer remain pay-per-use regardless of plan.
