Method
usage = top-k × chunk + prompt + output reserve.
Top-k context budget
Check how many tokens retrieved chunks consume before the model call.
· Herramienta local⚡ Instant local calculation◎ No sign-up↔ ES / EN
Inputs
Change assumptions
How to use it
Use the result as a transparent planning estimate. Change one assumption at a time, compare scenarios and move to Toolkit when you need system-level simulation.
Limits
Reranking may send fewer chunks to the model than initially retrieved.
◉
Local by design
Your inputs are processed in this browser. This utility does not need an account, database or AI API.