Search Ctrl K
Home / Tools
THE FEUTEX TOOLKIT

LLM Prefill and Decode Latency

Estimate request latency from measured prompt throughput and output-token rate.

AI Infrastructure workspace● Runs in your browser
01 Your input
02 Your result
View or copy raw result data
FormulaEstimated first-token time = queue+input/prefill+1/decode; Estimated full response time = queue+input/prefill+output/decode

Made for real tasks

  • Estimate request latency from measured prompt throughput and output-token rate.
Explore step-by-step workflows →

Worked example

Inputs

Prompt tokens
2000 tokens
Output tokens
300 tokens
Prefill throughput
10000 tokens/s
Decode rate
50 tokens/s
Queue and network time
0.1 s

Expected results

Estimated first-token time
0.32 s
Estimated full response time
6.3 s

Load the example in the workspace to reproduce these results. Display rounding does not change the underlying calculation.

How to use this tool

  1. Enter prompt tokens, output tokens, prefill throughput, decode rate, queue and network time in the displayed units.
  2. Calculate the scenario, then compare it with measurements before changing a system.

Try this example

{
  "input": 2000,
  "output": 300,
  "prefill": 10000,
  "decode": 50,
  "queue": 0.1
}

Accuracy and scope

Constant measured rates; excludes dynamic batching, speculative decoding and token-rate variation.

Is my data uploaded?

This tool runs locally in your browser. Input and output are not sent to a processing server. Copy or download your result before leaving the page.