Anthropic's prompt caching feature lets the model reuse cached KV state for repeated context. For a tutor product with a long, static system prompt and a per-problem context that persists across many turns, this is a large cost lever.
We cache every system prompt in the codebase with cache_control: ephemeral. Tutor, hint, reveal, grader, classifier, leak guard, recall rephrase, normal assistant — all cached. Cache TTL is around five minutes, which is longer than most single-session engagement.
We additionally cache the per-problem context — the problem statement plus the student's confidence tag — as a second block. This means multiple turns on the same problem (attempt → tutor reply → student re-attempt → tutor reply) share the same cache.
What we will not do is quote you a saving. Every system prompt in the codebase is cached and every per-problem context is a second cached block — that is verifiable in the repository — but we are pre-launch, and a hit rate measured over a handful of sessions is not a number anyone should plan around. Anthropic publishes the mechanism and its pricing; the saving on your traffic is theirs to compute, not ours to assert. When we have volume, the measurement will appear here with its method.
The engineering cost was two lines per prompt call. This was one of the highest-leverage engineering weeks in the project.
Written by Brainback Engineering. Published Jul 16, 2026. Filed under Method.