Method
retrieval latency = sum of all components.
RAG retrieval latency
Break down retrieval latency before the LLM 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
Measure p50 and p95 separately; adding averages does not model queueing.
◉
Local by design
Your inputs are processed in this browser. This utility does not need an account, database or AI API.