tokevenThere is no baseline
There is no baseline period, no 30-day observation window, and nothing frozen. You do not have to wait to start saving, and we never bill you against a snapshot of what you used to spend.
Per-call deltas
Savings are measured one call at a time, at the moment the call happens.
When you accept a recommendation - moving a call to a cheaper model, or restructuring a prompt so it hits the cache - the wrapper records which model you would otherwise have used. The server then prices both sides of that single call and stores the difference:
- What the call actually cost, on the model you sent it to
- What the same call would have cost on the model you moved away from
- The delta - your verified saving for that call
Your savings for a month are the sum of those per-call deltas. Nothing is estimated across a period, and a call you did not change contributes nothing either way.
Both sides are priced live
Both figures use the token prices in force at the time of the call, not prices captured when you signed up. That is the point of having no frozen baseline: when a provider cuts its prices, the delta on that call shrinks, and our fee falls with it automatically. A frozen baseline would keep billing you for a saving the market had already handed you.
The fee
We keep 10% of verified savings. You keep 90%. If a month produces no savings, there is no fee - there is nothing else to pay.
Auditability
Every savings claim links to:
- •The specific call that was optimized
- •The model change or caching improvement behind it
- •Both priced amounts and the resulting delta
- •The token prices used, and when they applied
Disputes
If you disagree with a savings calculation, contact us. We refund promptly and investigate. Our goal is zero friction for honest clients.