Input vs output tokens

Two meters, two prices

Every API call has two token counts — what you send in (input) and what Claude sends back (output). They're priced differently, and output costs more per token.

A prompt that generates a 2,000-word essay costs significantly more than one that generates a 3-sentence summary — even if both prompts are the same length. Controlling output length is one of the fastest ways to cut costs.

Always specify length

Always tell Claude how long you want the response. "In 2–3 sentences" or "as a bullet list under 100 words" saves real money at scale.