AI Daily Grind // AI news, freshly ground daily

Learn · Models and how they work

What is Token?

Illustration: Token
Simply put, like I'm 5 years oldA token is like one small piece of a chopped carrot. The AI does not eat a whole sentence at once: it cuts the words into little pieces and reads them one by one.

The slightly longer version

A token is a small chunk of text that an AI model reads or writes. It can be a whole short word, a piece of a longer word, a space or a punctuation mark. Before a model reads your message, a tool cuts the text into tokens and turns each one into a number. The model then writes its answer one token at a time, guessing the next piece again and again.

Tokens are also how AI is measured and sold. A "context" is how much text the model can keep in view at once, counted in tokens. So a "1-million-token context" means the model can look at a very long text in one go. Many companies also count your usage, and your bill, in tokens.

How a model uses tokens

  1. You write textYou type a question or paste a document.
  2. Text is cutA tool splits the text into small chunks called tokens.
  3. Tokens become numbersEach token gets a number the model can work with.
  4. Model guesses nextThe model writes its answer one token at a time.
  5. Tokens are countedThe count sets the context limit and often your bill.
Why it matters

In today's news, one person ran a model with a 1-million-token context on a gaming card at home, while big labs ration access because agents eat so many tokens. And a research paper says the first few tokens of a model's "thinking" may decide how well we can watch it. Knowing what a token is helps you read these stories, and understand why your AI plan has limits.

What to watch out for

There is no fixed size: each company cuts text in its own way, so "one token" from one model is not the same as from another. Languages other than English, like French, often need more tokens for the same idea, so they can cost more. And a big context number does not prove the model really uses all that text well; that depends on the model and is often not clear.

Where it showed up

This notion came up in More agents, fewer eyes on them. More lessons: Learn.

Sources

Written on 2026-11-02. This explanation was accurate on that day. AI changes fast, so check the sources and the date before you rely on it.

Get The AI Daily Grind by email, every morning, in simple words.