THE FEUTEX TOOLKIT

Speculative Decoding Token Yield

Estimate accepted draft-prefix tokens for a fixed draft length and independent acceptance rate.

AI Infrastructure workspace● Runs in your browser

01 Your input

Draft tokens per proposaltokensPer-token conditional acceptance%

02 Your result

View or copy raw result data

Result data

Formula

Expected accepted draft tokens = p===100?k:(p/100)*(1-(p/100)**k)/(1-p/100); Expected rejected draft work = k-(p===100?k:(p/100)*(1-(p/100)**k)/(1-p/100))

Made for real tasks

Estimate accepted draft-prefix tokens for a fixed draft length and independent acceptance rate.

Explore step-by-step workflows →

Worked example

Inputs

Draft tokens per proposal: 4 tokens
Per-token conditional acceptance: 80 %

Expected results

Expected accepted draft tokens: 2.3616 tokens
Expected rejected draft work: 1.6384 tokens

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

How to use this tool

Enter draft tokens per proposal, per-token conditional acceptance in the displayed units.
Calculate the scenario, then compare it with measurements before changing a system.

Try this example

{
  "k": 4,
  "p": 80
}

Accuracy and scope

Accepted prefix model only; not end-to-end speedup, which depends on verifier and draft costs.

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.