Skip to content
← Inference System Design / 20
hardNewAnthropicOpenAIFireworks AI

Price a prompt-cache hit from the server side: what it saves, what holding a prefix costs, why long retention costs more

A cached input token is cheaper for the customer, but the discount has to come from somewhere. The answer prices what a hit saves in replica time, what keeping the prefix costs in each memory tier, and shows why a retention guarantee changes the arithmetic that opportunistic caching never had to do.

Updated Nov 2026 · Worked answers on inference mechanisms, with calculations, implementation details, and follow-up questions.

A cached input token is cheaper for the customer, but the discount has to come from somewhere. The answer prices what a hit saves in replica time, what keeping the prefix costs in each memory tier, and shows why a retention guarantee changes the arithmetic that opportunistic caching never had to do.

10 answers per topic instead of 5, plus saved progress and bookmarks · costs nothingor unlock all 213 remaining answers · $25

Every worked answer and derivation in all 12 topics, every concept page, the full practice-test bank, and every code drill’s edge-case tests and reference solution.

6 months · One payment · No auto-renewal

UP NEXT ON YOUR JOURNEY
Next in this trackShould small inference services share a GPU through MIG, MPS or time slicing? What isolation do they get?Popular right nowWhy is batch-1 decode on a 70B model bandwidth-bound, and what is its latency floor on two H100s?Popular right nowDecode attention intensity is fixed at the query-to-KV head ratio. Derive that, then tell me what MLA does to it.
DISCUSSION · 0

Nobody has commented on this answer. Add a correction or another way to work it.