Read the token bill — what your LLM feature actually costs (step 2/9) · production tradeoffs

Four prompt shapes, all running on Claude Sonnet 4.6 ($3/1M input, $15/1M output). Which costs the MOST per call?

Be careful — output is 5× the price of input, so token count alone doesn't tell you the answer.

Four prompt shapes, all running on Claude Sonnet 4.6 ($3/1M input, $15/1M output). Which costs the MOST per call?

Be careful — output is 5× the price of input, so token count alone doesn't tell you the answer.

Read the token bill — what your LLM feature actually costs — step 2 of 9