Output is most of this bill
When output is most of the bill, shortening the prompt barely moves the total.
Output tokens cost five to six times more than input tokens on most models. A workload that generates long responses can spend the large majority of its budget on completions, in which case prompt optimisation has a small ceiling no matter how thorough it is.
This is worth knowing before you spend a day compressing a system prompt. If output is 80% of the bill, halving the prompt changes the total by ten percent.
The levers that matter for output are different ones: cap `max_tokens` to something your product can actually display, ask for the shortest useful form rather than an essay, avoid instructing the model to restate the question or explain its reasoning when you will not show it, and use a schema so it emits fields instead of prose.
Reasoning models add a wrinkle: thinking tokens are billed as output even though nobody reads them. If you are on a reasoning model for a task that does not need it, that is the first thing to check.
2,000-token prompt, 3,000-token response. Prompt is 12% of the bill.Cap the response at 400 tokens and the bill falls by more than half.Check your own prompt
The analyser checks this pattern along with the other 25, prices each finding against your request volume, and hands back a rewritten prompt. It runs in your browser — nothing is uploaded.
Run the analyserMore on model choice
- A newer model at a lower priceSome model upgrades cost less on both input and output than the model you're currently using.
- Your prompt crossed a pricing cliffSome models double their input price above a prompt-size threshold.
- This model counts your prompt differentlyThe same text is not the same number of tokens on every model. Upgrades can raise your bill silently.
- Running out of roomA prompt filling most of the context window leaves nothing for the conversation.