GardenBench introduces 11 test suites designed to track daily fluctuations in LLM accuracy, latency, and run costs. The framework identifies performance degradation or unexpected changes occurring after model provider updates or deployments.
HOW THIS AFFECTS YOU
●
builderYou can use this to detect when model provider updates unexpectedly alter your application's performance or cost structure.