I would like to ask which one of the
is better to use after I finished my texture rendering. Is there any difference performance-wise?
These are two fundamentally different things.
There is a little bit of validation overhead associated with binding any OpenGL resource (GLSL programs and FBOs tend to have the most complicated validation). In general, changing an object binding tends to be more expensive than changing one or two states associated with a bound object.
Therefore, unnecessarily changing the bound texture is inadvisable.
If your intention by binding 0 was simply to "disable" a texture, you would be better off setting a uniform value to communicate this intent to a shader. Uniforms are extremely cheap to set (assuming you do not do something stupid like query their location everytime you set them). Changing a uniform is orders of magnitude quicker than binding a different texture or even worse a different GLSL program.
If you are using fixed-function OpenGL:
If you are using programmable OpenGL: