How AI Usage works
AI Usage measures the amount and type of AI work FynOps performs for your company. It replaces one-size-fits-all call counting with a meter that reflects model cost.
One unit
One AI Usage Unit represents $0.009 of provider list-price-equivalent inference. It is not a fixed token bundle. Input, output, cached input, reasoning, model choice, and supported media can carry different weights.
Examples
Fractional use
A short, efficient request can use less than one unit. Your billing page shows two decimals while enforcement keeps full precision.
One-unit reference
On Gemini 3.1 Pro Preview, 1,500 input tokens plus 500 output tokens equal exactly one unit at the pinned rate.
Larger work
10,000 input tokens are about 2.22 units at the reference rate. 10,000 output tokens are about 13.33 units because generated output costs more.
Included each billing period
Free
150
unitsStarter
450
unitsGrowth
1,500
unitsScale
3,000
unitsUsage resets at the billing-period boundary shown in the app. Free blocks new AI work at its limit. Paid plans can continue into overage at the rate shown in Billing. Provider retries owned by FynOps do not add customer units, but a new send, regeneration, agent turn, or automation model turn is new work.
What does not use AI Usage
Searching or opening chat history, deterministic calculations, data formatting, graph queries, and tools that do not invoke a model do not use AI Usage by themselves.
AI Usage and Platform Credits
AI Usage is a service allowance. Platform Credits are monetary funds that can reduce eligible charges across the platform. They are separate balances. Using AI Usage does not spend a Platform Credit directly; any eligible credit is applied after charges are posted.
Questions about your current period? Open Settings and Billing in the app.