The number of bits of a single chip microcomputer (reprinted)

It refers to the number of bits of data that the microcontroller processes at one time.
 
Generally, 8-bit microcontrollers are low-end microcontrollers, 16-bit microcontrollers are mid-range microcontrollers, and 32-bit microcontrollers are high-end microcontrollers.

If the bus width is the same as the data width processed by the CPU at one time, then this width is what is called the number of bits in the microcontroller.
If the bus width is different from the data width processed by the CPU at one time:
1) The bus width is smaller than the data width processed by the CPU at one time, then the number of bits in the microcontroller is defined by the data width of the CPU, but it is called quasi-bit. For example, the famous Intel  8088 has a 16-bit CPU but an 8-bit bus, so it is quasi-16-bit.
2) The bus width is smaller than the data width processed by the CPU at one time, then the number of bits in the microcontroller is defined by the data width of the CPU.

Bit width does not refer to the bus width or the memory width. For example, the address bus of the 51 microcontroller is 16 bits, but it is an 8-bit machine. The memory of ARM is also 8 bits, but it is a 32-bit machine. It refers to the width of the data processed by the CPU, that is, the CPU's data throughput at one time. For example, the same instruction: MOV  R0  R2
in the 51 microcontroller, R0 and R2 are both 8 bits, so the 51 CPU can only process 8 bits of data at a time.
In ARM, R0 and R2 are 32 bits, so the ARM CPU can process 32 bits of data at a time. This is the difference.