xAI's $0.05/min Promise: The Hidden Cost Reality for Agencies
xAI just launched their Voice Agent Builder with aggressive $0.05/min pricing. But what they don't tell you is that actual all-in costs reach $0.12–$0.25/min. Here's the breakdown.
The xAI Announcement
On July 1, 2026, xAI announced their Voice Agent Builder, pricing it at an eye-watering $0.05/minute for platform use. On the surface, it undercuts Vapi ($0.05 base) and Retell ($0.07–$0.18/min) by claiming to be the cheapest option on the market.
Within 48 hours, every AI voice agency builder in Discord and Skool was debating whether to switch. "xAI at $0.05 is a no-brainer," the thinking goes. "That's half of what I'm paying now."
Here's the problem: that $0.05 number is a trap. And if your agency falls for it, your margins will suffer immediately.
Red flag: Any platform quoting per-minute cost that doesn't include STT, LLM, and TTS is being deceptive. It's not misleading by accident. It's by design.
Why This Matters for AI Voice Agencies
Agencies already got burned once. Voicerr raised prices 7–10x in early 2026. GoHighLevel bundled voice as an add-on. Synthflow priced agencies out of the market.
The pattern is the same every time: headline pricing that excludes costs, followed by invoice shock 30–60 days in.
If you build your client pricing around xAI's $0.05/min rate, and your actual costs turn out to be $0.15–$0.20/min, your margins disappear. You're running a charity, not an agency.
The key insight: "Per-minute" pricing without full transparency is a red flag. Full stop. This is not a platform feature; it's a marketing tactic.
xAI's $0.05: What's Actually Included?
xAI's $0.05/min breaks down like this:
xAI Platform Fee: $0.05/min
What it covers: API calls, routing, session management, basic logging.
What it does NOT cover: Speech-to-text, large language model inference, text-to-speech, telephony.
Speech-to-Text (STT): +$0.02–$0.04/min
xAI uses third-party STT. That cost passes through to you. Whisper-quality STT is $0.02–$0.04/min.
Large Language Model (LLM): +$0.03–$0.06/min
xAI runs on Grok, but Grok inference is expensive. Expect $0.03–$0.06/min for model API calls, depending on token use and model tier.
Text-to-Speech (TTS): +$0.02–$0.05/min
Natural-sounding voice (Google Cloud TTS, Azure TTS, ElevenLabs) runs $0.02–$0.05/min. Cheap TTS (generic) is $0.01/min but hurts agent quality.
Telephony (Twilio, CallKit, etc.): +$0.01–$0.03/min
Inbound and outbound calls via Twilio or equivalent cost $0.01–$0.03/min. This is unavoidable.
Total All-In Cost
Platform: $0.05
STT: $0.03
LLM: $0.04
TTS: $0.03
Telephony: $0.02
= $0.17/min minimum
With premium TTS and LLM: $0.25/min
Stop chasing per-minute platforms. Move to transparent pricing.
Hermes gives you predictable all-in costs. Voice agents, CRM, campaigns. One platform. Your brand. From $149/month.