Token
A token is one small piece of text the model reads before it can answer.
Plain English
A token is one small piece of text the model reads while it is trying to understand your request. Think of a sentence getting broken into little pieces the model can count, track, and work through. Short requests use fewer pieces. Long notes, manuals, and chats use more.
Why it matters here
If you want AI to read long shift notes, alarm history, and setup instructions together, token usage becomes a real limit. That affects how much context you can send at once.
Picture it like this
Use simple language and analogies before technical phrasing takes over.
Picture a sentence being snapped into small text pieces before the model works on it. The model does not hold one giant thought bubble. It reads piece by piece, then builds the answer from those pieces.
This is a simple picture, not the exact tokenizer output. Real token splits vary by model, punctuation, and wording.
Analogy
The model keeps count of many small text pieces while it reads, reasons, and answers.
Like checking whether a message is short enough to scan quickly or long enough that it needs to be broken into smaller parts to stay clear.
Examples
Related links
Use these pages to move from definition into adjacent workflows or broader context.
Featured term pages
These pages keep the deeper glossary content crawlable instead of leaving everything buried inside an accordion list.