English
Deutsch
PolskiCheck token cost, retry attempts and peak usage against a monthly cost budget per customer.
All monetary amounts use the same currency and exclude VAT. Example values are assumptions, not provider prices. Inputs and calculations stay in the browser.
Calculation model and assumptions
Cost per attempt = input tokens / one million × weighted input price + output tokens / one million × output price + additional cost. Weighted input price uses actual cache reads only. Cost per original request = attempt cost × (1 + additional retry rate). Usable budget = budget × (1 − reserve). The limit is original requests rounded down to fit that budget. The table shows normal, double and peak usage. Prices are hypothetical inputs, not current provider tariffs. Cache-write premiums, multimodal units and external tools belong in additional cost or a separate calculation.
More SaaS tools
Sources and definitions
FAQ
Are inputs transmitted?
The calculator processes inputs locally in the browser. CSV files are also generated locally. The page does not store scenario inputs on the server.
What are the model limitations?
The calculator does not enforce technical limits. Tokens and retries are averages. A zero-cost scenario has no finite cost limit; actual rate limits, concurrency and abuse protection require separate implementation.