Matmul
Also known as: Matrix Multiplication, Matrix Multiply
A matmul (matrix multiplication) multiplies two matrices (or a matrix and a vector) with the usual linear-algebra rule: entry $(i,j)$ of $AB$ is the dot product of row $i$ of $A$ with column $j$ of $B$.
In attention code you see matmuls constantly—$QK^\top$ for scores, attention weights times $V$ for the context, and the $W_Q, W_K, W_V$ projections. Libraries expose it as @ / torch.matmul / np.matmul; GPUs are built to run large matmuls fast.