$ cache-savings

Pick a scenario. Know if prompt caching is worth it — in 30 seconds. Pricing updated 2026-06-01.

# scenario

Daily scripts & tooling · 5K tokens × 20 turns/day · 70% cache hit rate

# model

# workload

Turns per day turns/day 20

Cache hit rate % reused from cache 70%

Cache TTL write cost

Advanced (context size, utilization, etc.)

Context size tokens/turn 5K

Turns per conversation turns/conv 8

Context utilization utilization 70%

Output / input ratio output ratio 40%

estimated monthly savings

Caching saves 0% vs no cache · ~$0/yr

Without cache$0/mo

With cache$0/mo

>>…