Create mock data with specific distribution for column values in Pandas

Here is the existing code

import pandas as pd
import random

random_seed = 1234
random.seed(random_seed)

def create_rows(num=1):
    output = [{ "ID":random.randint(1000,2000),
                "CLASS":random.choice(['A', 'B', 'C']),
                } for x in range(num)
               ]
    return output

df = pd.DataFrame(create_rows(5000))

print (df.head())

How can I create the CLASS column in such a way that 'A' appears 50% of the time, while 'B' and 'C' each appears 25% of the time?