Output tokens
Output tokens are the text a model generates: explanations, code, and — with reasoning models — internal thinking that is billed even when it is not shown. They cost several times more per token than input, and they are capped per response.
The price asymmetry shapes good practice in a non-obvious way. Asking a model to read a lot and write a little is cheap. Asking it to regenerate an entire file to change three lines is expensive, and it is what happens when a request is phrased as "rewrite this file with X" instead of "change X".
The per-response cap is a separate constraint. Past it, generation stops mid-token regardless of how much more there was to say.
What this means in practice
Ask for diffs or targeted edits rather than whole-file rewrites when the change is small — cheaper, faster, and far easier to review. When output truncates, split the request by file rather than asking it to continue.
Prompts for the tools this applies to
Related terms
- Input tokens
Input tokens are the units of text a model reads: your prompt, the files supplied, the system prompt, and the conversation so far.
- Token limit
A token limit is a cap on how much text can be processed in a request or a period.
- Context window
The context window is the maximum amount of text a model can consider at once, measured in tokens and covering everything in the request: system prompt, your instructions, the files supplied, prior conversation, and the response being generated..