Glossary term
TOPS (Tera Operations Per Second)
An AI inference performance metric counting trillions of operations per second; common in edge and mobile AI specs.
TOPS (Tera Operations Per Second) is an AI performance metric counting trillions of operations per second, used most often for inference and especially in edge and mobile AI. It is the integer-math counterpart to FLOPS, which measures floating-point throughput, and headline TOPS figures depend on the precision assumed. TOPS is the spec marketers cite for NPUs in laptops and phones. See Custom Silicon 101.