Parameters
The internal numbers (weights) a neural network adjusts during training — its learned knowledge, stored as billions of values. Model size is measured in parameters: an 8B model has 8 billion of them. More parameters generally means more capability but slower, costlier inference. For creators, parameter count is a rough quality signal when comparing models, not a guarantee — architecture and training data matter just as much. Example: a smaller, well-trained image model can beat a bloated one, which is why benchmarks and your own test prompts beat spec sheets every time.