Cache write tokens

Cache write tokens are the tokens charged the first time content is stored in the prompt cache, at a small premium over the input price.

Cache write tokens are what you pay to put content into the prompt cache the first time. They cost a little more than a plain input token: 1.25 times the input price for the 5-minute cache, or 2 times for the 1-hour cache. After that, reading the content back is cheap.

Because of that premium, caching pays off once the cached content is read back: after the first read for the 5-minute window, or the second for the 1-hour window. The cost calculator has a 5-minute and 1-hour toggle, so you can see which window is worth it for a given run.

Checked against 5-min write 1.25x, 1-hour write 2x input (2026-10-23) .

All terms →