Home › AI glossary › FLOPs
What is FLOPs?
FLOPs count the math steps a computer does, and AI people use them to measure how much work a model needs.
- Category Hardware
- Also called FLOP, floating-point operations
- Updated 20 Oct 2026
FLOPs, explained simply
FLOP stands for floating-point operation. That is one small math step with decimal numbers.
FLOPs with a small s means a count of those steps. FLOPS with a capital S means steps per second, which is a speed.
Training a big model takes an enormous number of them. People use FLOPs to compare how much work different models took.
On ITG Now, our AI research page and daily news stories sometimes use FLOPs to describe how much work it takes to train a model.
Related terms
ComputeGPUTrainingParametersAI chip
Questions people ask about FLOPs
FLOPs vs FLOPS: what is the difference?
FLOPs is a total count of math steps. FLOPS is how many steps a chip can do each second.
Why do people measure AI in FLOPs?
It gives a fair way to compare the work done, no matter which chips were used.
Do I need to know FLOPs to use AI?
No. It is a measure mostly used by researchers and chip makers.