CacheCraft
Benchmark that replays recorded request traces through candidate semantic caching policies.
Desktopfreeglobal
CacheCraft is a free, dependency-free benchmark that replays a recorded trace of real requests through candidate semantic caching policies, measuring the actual latency saved and quality risk accepted by each one. It is aimed at engineers operating LLM-backed applications who want to choose a caching strategy based on measurement against their own traffic rather than guesswork.
Categories
AI infrastructure toolsbenchmarking
Something wrong with this listing — dead link, not a real product, wrong info?

