What the commission is, and why.
A percentage added to what the provider charges for each AI call, on usage beyond what your plan includes. It is the same for every model and falls as the plan grows. Here is how it is worked out, what it pays for, and why AI usage is priced this way.
- Included usage, spent first
- -$50.00
- Beyond it, at provider rate
- $450.00
- Commission, 8%
- +$36.00
- Usage billed for the month
- $486.00
On top of the plan price. Enterprise is negotiated, down to 0%.
Three steps, the same on every plan.
The commission never touches the subscription or the usage your plan includes. It is a percentage on the usage beyond that, and nothing else.
- 01
Included usage goes first
Every paid plan includes AI usage each month - $10 on Starter, $50 on Professional, $100 on Business, and up to $1,000 on Enterprise. It is spent first, at the provider's rate, with no commission on it. Free starts with a one-time $5.
- 02
Then each call is metered at the provider's rate
Beyond the included amount, every call is metered at what the provider charges for that model and that call. Rates follow provider pricing and are shown in the workspace before anyone sends.
- 03
Your plan's percentage goes on top
The commission is one percentage per plan, the same for every model, maker and kind of output. Usage beyond the included amount is billed in arrears, separately from the subscription.
| Plan | Included usage | Commission | $1,000 beyond it is billed at |
|---|---|---|---|
| Free | $5 one time | 20% | $1,200 |
| Starter | $10 / month | 10% | $1,100 |
| Professional | $50 / month | 8% | $1,080 |
| Business | $100 / month | 5% | $1,050 |
| Enterprise | Up to $1,000 / month | Down to 0% | Negotiated |
The platform around every call.
The provider's rate pays for the model. The commission pays for everything that makes the call usable inside a company - and it is the same whichever model you choose.
One published percentage, instead of a price you cannot check.
There were other ways to price this: a credit currency, marked-up model prices, or a large AI allowance bundled into every seat. Each of them hides something. This is why we chose the one that does not.
- You see the real price The provider's rate is the provider's rate, and the commission is one published number on top. There is no credit currency in between and no private price list to decode.
- Neutral about models The same percentage applies to every model, so nothing in how we are paid pulls you towards one model over another. Routing goes to the best model for the job.
- It follows the work, not the headcount Seats do not carry a bundled AI allowance that light users pay for and heavy users run out of. Usage is priced as it happens, after the amount every paid plan includes.
- It thins as you commit 20% on Free, 10% on Starter, 8% on Professional, 5% on Business, and negotiable down to 0% on Enterprise, as part of the contract.
Back to the plans or run the TCO calculator to see usage next to the subscription and the time around it.
What finance asks about the commission.
Is there a commission on the usage my plan includes?
No. The included usage is spent first, at the provider's rate. The commission applies only to usage beyond it.
Is the commission charged on the subscription or on seats?
No. It applies to AI usage only. The plan price and any add-on seats are exactly what the pricing page says they are.
Is it the same for every model?
Yes. It is one percentage per plan, whichever model you pick and whatever it produces - text, images, video or audio. What differs between models is the provider's own rate.
Where does the provider's rate come from, and what if it changes?
It is what the model's provider charges for that call. Provider pricing can change, and changes are reflected in the workspace first, where each model's rate is shown before you send.
How do we see what the usage comes to?
Usage Statistics breaks AI usage down by person, by team, by model and over time, so it can be followed during the month and recharged internally afterwards.
Why not build the margin into the model prices instead?
Because then nobody could tell what a model really costs. Keeping the provider's rate as it is and adding one published percentage means the price of any call can be checked against the provider's own price list.
Can the commission go to zero?
On Enterprise it is negotiated as part of the contract, down to 0%. That is also where the private deployment options sit: your own provider keys on dedicated infrastructure, privately hosted open-weight models, or a self-hosted install.
Prices are in USD and exclude any applicable tax. The examples are illustrative; the rate for each model is shown in the workspace before you send. Last reviewed 5 October 2026.
The price of every call, before you send.
Start free with $5 of AI usage on any model. No card and no procurement marathon - just the real rates, on real work.