Tokens (the AI kind, not design tokens)
The small chunks of text an AI reads and writes, and what your usage is measured in. Not to be confused with design tokens, which are reusable style values.
What it is
In AI, a token is a small chunk of text. Roughly, one token is about four characters, so a short word is one token and a long one is two or three. The model reads tokens and writes tokens, and your usage and cost are counted in tokens. Everything you send and everything it returns is measured this way.
It is the word count on a translation invoice: nothing to do with the tokens in your Figma file. The model chops every message into small chunks of text and charges per chunk, the way a telegram charged per word. Design tokens name colors and spacing; these tokens are just the syllables the AI reads and writes.
The naming clash designers hit
Designers already know “tokens” as design tokens: reusable values like a color or a spacing step. These are a completely different thing. When an AI tool talks about tokens, it means these text chunks and usage, not your color variables. Same word, two worlds.
Why this matters for designers
Tokens are why design work burns AI allowance fast. A pasted screenshot or a full design-system file is a large block of tokens, and the model re-reads them every turn. Knowing this changes how you work: send what the task needs, not everything, and your sessions get cheaper and sharper.