Computer Networks – Basic Networking Question – Why the maximum data size of the BYTES ip 65535 datagram and not 65535 bits?

I have recently started computer networks with some understanding of algorithms and discrete mathematics. I am completely new. Please elaborate below –

ip datagram data size = 16bits. So with that $ 2 ^ {16} -1 = $ 65536 16-bit models can be generated.
My doubt:

  1. Do we say that each 16-bit model corresponds to an IP datagram or to all these models as to an IP datagram?
  2. If I consider all 65536 models as the maximum data size, then the maximum size of the ip datagram $ = $ 65536 size of the bosses$ = 65536 * $ 16 parts $ = 65536 * $ 16/8 bytes $ = 65536 * $ 2 bytes