Token usage is the amount of text an AI processes, counted in small pieces called tokens. Some AI services use the number of input and output tokens to calculate how much a request costs.
For example, a long set of instructions and files may use more input tokens than a short question. A very long answer may also use more output tokens.
The exact cost depends on the service, model, and current pricing rules. Check the provider's current pricing page instead of relying on an old estimate, and avoid sending extra information the task does not need.