Prompt Caching Savings Calculator for Claude
See how much prompt caching saves on your Claude API bill. Enter your cached prompt size, request volume and hit rate to get cost with and without caching and your break-even hit rate. Free.
Your workload
How to use the Prompt Caching Calculator
- Pick a Claude model to fill in its official prices, or choose Custom prices and type your own.
- Choose the cache lifetime: 5 minutes (writes cost 1.25x) or 1 hour (writes cost 2x).
- Enter the size of the cached prefix, the new input and the output per request.
- Enter how many requests you send and the share of them that hit the cache.
- Compare the cost with and without caching, and check the break-even hit rate.
Frequently asked questions
How much does prompt caching save on Claude?
It depends on the model and your hit rate. Anthropic charges 1.25x the base input price to write a 5-minute cache entry and 2x for a 1-hour entry, while cache reads cost a fraction of base input. The savings apply to the cached prefix only, not to new input or output.
Are cache read prices the same for every Claude model?
No. Anthropic’s pricing page lists 0.1x base input as the standard cache hit price, but 0.05x for Claude Opus 5.5 and Claude Sonnet 5.5, and 0.025x for Claude Fable 5.1. This calculator fills in the correct multiplier per model.
What is the break-even cache hit rate?
It is the share of requests that must hit the cache for caching to cost less than sending the full prompt every time. It equals (write multiplier – 1) divided by (write multiplier – read multiplier). For a 5-minute cache on Opus 5.5 that is about 21 percent.
What counts as a cache hit here?
Any request that reads the cached prefix instead of writing it. The calculator treats the rest of your requests as cache writes, which is a simple model. Real hit rates depend on how often you reuse the same prefix inside the cache lifetime.
Where do the prices come from?
Anthropic’s official pricing page, last checked on 2026-11-01. Prices change, so confirm them on the pricing page before you budget, or switch to Custom prices.
Does this work for OpenAI or Gemini?
Choose Custom prices and enter that provider’s base input price, read multiplier and write multiplier. Caching rules differ by provider, so check their documentation.