figure.add_subplot() vs pyplot.subplot()

If you need a reference to ax for later use:

ax = fig.add_subplot(111)

gives you one while with:

plt.subplot(111)

you would need to do something like:

ax = plt.gca()

Likewise, if want to manipulate the figure later:

fig = plt.figure()

gives you a reference right away instead of:

fig = plt.gcf()

Getting explicit references is even more useful if you work with multiple subplots of figures. Compare:

figures = [plt.figure() for _ in range(5)]

with:

figures = []
for _ in range(5):
    plt.figure()
    figures.append(plt.gcf())

pyplot.subplot is wrapper of Figure.add_subplot with a difference in behavior. Creating a subplot with pyplot.subplot will delete any pre-existing subplot that overlaps with it beyond sharing a boundary. If you do not want this behavior, use the Figure.add_subplot method or the pyplot.axes function instead. More

It's just like what Matplotlib document says, which has also been mentioned in @afruzan's answer.

matplotlib.pyplot.subplot: Creating a new Axes will delete any pre-existing Axes that overlaps with it beyond sharing a boundary. If you do not want this behavior, use the Figure.add_subplot method or the pyplot.axes function instead.

To make it more clear, here is an illustration:

• Use Figure.add_subplot:

import matplotlib.pyplot as plt
    
fig = plt.figure()
fig.add_subplot(131, facecolor='red')
fig.add_subplot(132, facecolor='green')
fig.add_subplot(133, facecolor='blue')
fig.add_subplot(231, facecolor='cyan')
plt.show()

• Use pyplot.subplot:

import matplotlib.pyplot as plt
    
fig = plt.figure()
fig.add_subplot(131, facecolor='red')
fig.add_subplot(132, facecolor='green')
fig.add_subplot(133, facecolor='blue')
plt.subplot(231, facecolor='cyan') # overlap with subplot generated by fig.add_subplot(131, facecolor='red'), so subplot generated by fig.add_subplot(131, facecolor='red') will be removed
plt.show()