What Is a Token?

When an AI model reads or generates text, it breaks the language down into smaller pieces called tokens. A token might be a whole word, part of a word, a number or even a piece of punctuation, depending on the text and the model being used.

So while we tend to think about how many words we’ve given an AI, the model is actually working in tokens. A short sentence might be broken into several of them before the AI processes what you’ve written and starts generating its response.

Why Do Tokens Matter?

Tokens matter because there is a limit to how much information an AI model can work with at once. The longer your conversation or document becomes, the more tokens you’re using, which is why you may sometimes come across token limits when working with large amounts of information.

Tokens can also affect how much AI costs to use. For businesses building AI into their own products or systems, providers often charge according to the number of tokens the AI processes and generates.

You don’t normally need to count tokens yourself when using an AI tool, but understanding what they are helps make sense of why AI has limits on how much information it can handle and why processing more information can cost more.