Key Takeaways
- Tokens are small pieces of text. Your prompt and the reply both get broken into them, and the count is what you are billed on.
- What you ask for drives the cost more than how often you ask. Depth is expensive, volume is cheap.
- Three routes: pay per use with no monthly fee, AI Employee Growth at $50 per month per location, Unlimited at $97 per month per location.
- Requests in AI Studio and AskAI are capped at a dollar, so the runaway bill people fear is not the real risk.
- "Usage included" means a rolling five hour window, not an open tap.
Nobody wants to put AI in front of a client and then get a bill they cannot explain. Emma from HighLevel published a short walkthrough on AI pricing that clears up the two things causing most of the confusion: what a token is, and what "usage included" actually covers. We run these numbers every week at On Point Tech Solutions in Los Gatos, so here is the same breakdown with the agency math attached. If you are shopping for a GHL consultant in the Bay Area, the AI usage dashboard is the first screen worth opening together.
Tokens Are the Unit, and Complexity Is the Cost
A token is a small piece of text. When you type a prompt, the model chops your words into these pieces to read them, then spends more of them writing the answer back. Emma's comparison is puzzle pieces, and it holds up.
The agency version: what you ask for matters more than how often you ask. A quick lookup on a contact is cheap. Asking the same system to research a market or build a page moves far more text both ways, and that shows up on the invoice. One heavy weekly report can cost more than hundreds of small automated replies. Volume is not the driver. Depth is.
The Three Plans and Who Each One Fits
Pay per use has no monthly subscription for AI features. You pick your model inside most HighLevel products, and each model carries its own rate, so a ChatGPT model and a Gemini model cost different amounts for the same job. Rates are in HighLevel's AI product pricing guide. Start here when you do not yet know what you will use.
AI Employee Growth at $50 per month per location covers most of what a small business touches in a normal week: a thousand Conversation AI responses, 100 Voice AI minutes, and 100 minutes of the Voice AI prompt optimizer. Unlimited at $97 per month per location stops metering Conversation AI, Voice AI, reviews and Content AI, and gives AskAI and AI Studio three times the usage allowance.
Two products behave differently on purpose. In AI Studio and AskAI you do not choose the model, and a request is capped so it cannot exceed a dollar. Most builds land in cents. That cap is worth knowing, because fear of an unbounded bill is what stops most owners from ever trying these tools.
The Five Hour Usage Window Is the Part People Miss
Several features on the paid plans are listed as "usage included," which reads like unlimited and is not. HighLevel runs a rolling five hour usage window. AI work is expensive to compute, so the window spreads demand out and keeps response times sane platform wide. That is standard practice, not a HighLevel quirk.
So a plan is not described by its price. It is described by how much you can get done inside a five hour stretch. HighLevel embedded a calculator in the pricing doc showing the projects you can run and the usage each burns, per plan. Spend ten minutes in it before you commit. They also shipped a wallet pricing view this week that puts your rates in one place, which pairs well with the usage dashboard once you are past sizing.
What a GHL Consultant in the Bay Area Checks on the Usage Dashboard
The AI usage dashboard in your agency level settings shows what each sub account spent and where. Three signals in it decide the plan. Which product the spend went to, because heavy Conversation AI use is a different call from heavy AI Studio use. Whether it is one location or many, because both paid plans are priced per location. And whether the pattern is bursty, because that is what the five hour window pushes back on, and where Unlimited earns its price.
Guessing costs money both ways. Overbuying wastes it, underbuying throttles the thing your client is paying you for.
Why This Matters for Bay Area and San Jose Businesses
Local service businesses here are the most exposed to a missed call. A roofer or a med spa in San Jose loses a job when the phone rings at 6pm and nobody picks up, and Voice AI and Conversation AI are the direct fix. The blocker has almost never been whether the tech works. It was that the owner could not tell whether it would cost $12 a month or $1,200.
Now the answer is checkable. One location on normal lead volume sits inside the $50 plan, which is less than a single missed roofing estimate here. You can prove that on a pricing page instead of asking a client to take your word for it. And if you resell these features, per location pricing makes your margin knowable per account rather than a mystery you eat.
Practical Steps
- Open the AI usage dashboard in your agency level settings and read the last 30 days per sub account, noting which products the spend went to rather than just the total.
- Run the usage calculator in HighLevel's AI product pricing guide against the projects you actually plan to build.
- Start quiet locations on pay per use. No monthly fee means an idle location costs nothing.
- Move a location to the $50 Growth plan once its Conversation AI or Voice AI volume is steady rather than occasional.
- Reserve the $97 Unlimited plan for bursty AI Studio or AskAI work, where the three times allowance is the point.
- Recheck in 30 days. A pricing call made on one week of data ages badly.
Watch the Full Breakdown
Final Thoughts
The confusion here was never really about the numbers. It was that the numbers came in units nobody uses day to day. Tokens, usage windows, allowances. Translate those into "what does one location cost me a month" and the decision takes ten minutes.
On Point Tech Solutions is a Go High Level consultant in Los Gatos serving San Jose and the greater Bay Area. We build websites, funnels and automation systems on HighLevel, and our clients own what we build. If you want a GHL consultant in the Bay Area to size your AI plan against real usage instead of a guess, we will do it for you or teach your team with an SOP. Start at optechsol.llc. Credit to Emma at HighLevel for the original walkthrough.