Concept AI Concepts

What Are Tokens? How AI Actually Reads Your Words

A plain-language visual explainer showing how AI breaks your words into tokens, why common words stay whole while rare ones shatter, and what that means for cost and long conversations.

Beginner

3 min

7 slides

6 views

July 30, 2026

Before an AI model can do anything with your message, it has to chop it up. Not into letters, and not quite into words either. It uses pieces called tokens, and once you can see them, a lot of confusing AI behavior starts to make sense: why long chats slow down, why costs climb, and why some words seem to trip the model up.

AI Concepts

What Are Tokens? How AI Actually Reads Your Words

A plain-language visual explainer showing how AI breaks your words into tokens, why common words stay whole while rare ones shatter, and what that means for cost and long conversations.

Your sentence gets chopped up first

You type a message. Before the model reads a single thing, your sentence is broken into small pieces called tokens. Everything the model does happens on those pieces, never on your sentence as a whole.

Technical: The step that splits text into tokens is called tokenization.

Metaphor: A sentence entering a machine and coming out as separate tiles

A token is not the same as a word

Some tokens are whole words. Some are half a word. Some are a single space or a comma. The pieces were chosen by how often they show up in text, not by grammar rules you learned in school.

Technical: Roughly speaking, 1,000 tokens is about 750 English words.

Metaphor: Puzzle pieces of uneven sizes forming one line

Common words stay whole. Rare ones shatter.

Everyday words travel as one tidy piece. Unusual names, technical terms, and invented words get broken into several fragments, because the model never saw them often enough to give them a piece of their own.

Technical: This is why unusual proper nouns often cost more tokens than common words of the same length.

Metaphor: One smooth tile beside a tile cracking into fragments

Spaces and punctuation are pieces too

The gaps between words, the commas, the line breaks, even the emoji all take up pieces. Nothing in your message is free. Formatting has a cost just like the writing does.

Technical: A leading space is usually bundled into the token that follows it.

Metaphor: Small dim tiles sitting between the bright ones

This is the thing you are paying for

Tools that charge by usage count tokens, not messages. They count what you send and what comes back. Two questions of the same length can cost different amounts depending on the words you picked.

Technical: Providers bill input tokens and output tokens separately, usually at different rates.

Metaphor: A meter counting tiles as they pass

Every long chat fills up the desk

A model has a fixed amount of room for the conversation. Each new message stacks more tiles onto the desk. When the desk fills, the earliest part of your chat slides off the edge, which is why a long conversation starts forgetting what you said at the beginning.

Technical: That fixed amount of room is the context window.

Metaphor: A desk stacked with tiles, the oldest sliding off the edge

Say more with fewer pieces

Once you can see the pieces, you write differently. Cut the throat-clearing. Paste only the part of the document that matters. Start a fresh chat when the topic changes. You get better answers and you pay for less.

Technical: Shorter, more specific context usually improves answer quality as well as cost.

Metaphor: A short bright row of tiles outperforming a long dim one

1 / 7

Want to go deeper?

Let's Work Together

Whether you need AI automation, strategic guidance, or want to explore what's possible, I'm here to help.

Get in Touch Learn About Me

What Are Tokens? How AI Actually Reads Your Words

What Are Tokens? How AI Actually Reads Your Words

Your sentence gets chopped up first

A token is not the same as a word

Common words stay whole. Rare ones shatter.

Spaces and punctuation are pieces too

This is the thing you are paying for

Every long chat fills up the desk

Say more with fewer pieces

More from Neural Canvas

What Is a Context Window? Why Long Chats Start Forgetting

What Is an AI Agent? The Difference Between Answering and Doing

What Is MCP? The Universal Adapter for AI Tools

Let's Work Together