← All questions
Are short AI prompts actually cheaper?
Often no. Vague prompts push the model to cover many interpretations with long output, and output tokens usually cost several times more than input tokens.
A typical vibe-coding loop is a tiny prompt, a huge output, a correction, another huge output. Costs rise, the context window fills and the model gets less precise.
The alternative is to put the effort into a clear definition up front so the model writes less and the first answer is the right one.