What is an alternative to creating spreadsheets with random numbers in Python ? using CSV

Question

When creating spreadsheets with random values in Python, I used a script containing the library openpyxl (see below). I want to increase the speed of creating random spreadsheets in python because in the current method it is very slow. How to get the same result by generating CSV first and separately?

The current problem is both in the time spent creating the files, how much RAM memory dedicated to mounting each spreadsheet. The spreadsheet, as currently configured, brings as a result approximately 91.2 MB (95,674 Kbytes) and this taking between 400 and 700 seconds (depending on other processes on the machine), as well as oscillating the RAM around 8GB in each cycle.

Question: Any suggestions (preferably in code) using CSV to supposedly increase the speed of generating spreadsheets containing random numbers? Follow the code (starting point) currently used

from openpyxl import Workbook
from openpyxl.compat import range
import random
import time

def gera_plan():  
    print('start script')
    t = time.time()
    numFile = random.randint(1,1001)
    dest_filename = 'file_workbook_' + str(numFile)+'.xlsx'

    wb = Workbook()
    intro = wb.active
    intro.title = 'data'
    intro['B2'] = 'Just Confidencial Data'

    ws = {'ws1','ws2','ws3','ws4','ws5','ws6'}
    #for work in range(1,10):
    for work in ws:
        t_name = 'data'+str(random.randint(1,1001))
        work = wb.create_sheet(title=t_name)
        #work = wb.active
        for row in range(1, 5000):
            #ws1.append(range(random.randint(100,300)))
            work.append(random.sample(range(1, 10000), 600))
    wb.save(filename = dest_filename)
    print('end script')
    elapsed = time.time() - t
    print('time %f seg' % elapsed) 

for ger in range(1,5):
    print('cicle:'+ str(ger))
    gera_plan()