Prompt caching savings calculator

If you send the same instructions or context with every request, caching lets you stop paying full price for that repeated part. Enter the bit that repeats and the bit that changes to see what caching would save - and how often you need to reuse it before it's worth it.

Prices verified against official provider pricing pages on 2026-06-29.

The formula

Without caching, you pay full price for the repeated part on every single request. With caching, you only pay full price the first time (plus a small fee to store it); after that the repeated part is much cheaper to read. The savings depend on how big the repeated part is and how often you reuse it before it expires.

Questions

How much can caching save?

It depends on how big your repeated text is and how often you reuse it. With long fixed instructions and lots of reuse, 50–90% off the cost of that text is common.

Can caching ever cost more?

Yes - some providers charge a little to store the text the first time. If you don't reuse it enough, or the repeated part is small, caching can lose money. The calculator shows the point where it starts paying off.

What does 'how often the cache gets reused' mean?

It's the share of requests where the repeated part is already stored and ready. The more your traffic reuses the exact same text, the higher it is - and the more you save.

Related reading and tools