A workload caches a 2K-token system prompt AND uses Batch API. On a ca — CCA Advanced Practice — Claude Cert Academy

A workload caches a 2K-token system prompt AND uses Batch API. On a cache hit, what is the approximate effective input price compared to standard synchronous pricing?

• ~50% (batch discount only — caching does not stack with batch)
• ~10% (cache hit only — batch discount does not apply to cached tokens)
• ~5% (cache hit AND batch discount stack)
• ~125% (cache write rate on every batch call)

Answer: ~5% (cache hit AND batch discount stack)

Cache reads stack with the batch discount. Cache hit gives ~10% of standard price; batch gives 50% off that already-discounted price. Combined: ~5% of the standard synchronous input rate. This stacking is the largest single cost lever for high-volume non-interactive workloads.

Continue to Claude Cert Academy