Benchmark (AI)
An AI benchmark is a standardized set of test tasks with predetermined evaluation metrics that measure how well language models or AI systems perform on specific capabilities such as reasoning, knowledge, coding, or safety.[1][2] Benchmarks enable reproducible, comparable assessment across different models and track progress in AI development.[1]
An AI benchmark is a standardized set of test tasks with predetermined evaluation metrics that measure how well language models or AI systems perform on specific capabilities such as reasoning, knowledge, coding, or safety.[1][2] Benchmarks enable reproducible, comparable assessment across different models and track progress in AI development.[1]