Five few-shot examples nearly doubled the cost of my short LLM prompt
DEV Community
Five few-shot examples nearly doubled the cost of my short LLM prompt
Every example you paste into a prompt is billed again on every call. In a short extraction task, five examples took the price per call from 1.00x to 1.92x on GPT-5.6 and to 2.04x on Claude and Gemini models. Keep the one example that fixes a real mistake, and let a JSON schema handle the format.
0 comments
No comments yet.