A team caches a 4K-token system prompt. Their first call writes the ca — CCA Advanced Practice — Claude Cert Academy

A team caches a 4K-token system prompt. Their first call writes the cache; subsequent calls within 5 minutes hit it. What is the rough billing impact on the cached tokens across two consecutive calls?

• First call: ~125% of input price. Second call: ~10% of input price.
• First call: 0%. Second call: 100% of input price.
• Both calls: ~50% of input price.
• First call: 100%. Second call: 0%.

Answer: First call: ~125% of input price. Second call: ~10% of input price.

Prompt caching charges ~125% of the input rate on a cache write (first time or after expiry) and ~10% on a cache read. The break-even is on the second call within the cache window — the average across the two calls becomes (125% + 10%) / 2 = 67%, already cheaper than 100%+100%.

Continue to Claude Cert Academy