If you want to setup your own cross-browser testing environment you could look at using Selenium Grid, have look here: https://www.lambdatest.com/blog/selenium-grid-4-tutorial-for-distributed-testing/

In the grid you install nodes (machines) that contain a fixed version of a browser, where you install a fixed version of the driver.

Maintaining your own grid is time consuming, probably more expensive than using something like BrowserStack which run the grid for you.

Asking the remote webdriver for a version of an OS/Browser will search the Grid nodes to find a suitable Node to run the test on. On the node the driver.exe will be started, etc...

desired_cap = {
 'os_version': '10',
 'resolution': '1920x1080',
 'browser': 'Chrome',
 'browser_version': '62.0',
 'os': 'Windows',
}
driver = webdriver.Remote(
    command_executor='https://your_grid/wd/hub',
    desired_capabilities=desired_cap)
driver.get("https://www.google.com")

If you want to setup your own cross-browser testing environment you could look at using Selenium Grid, have look here: https://www.lambdatest.com/blog/selenium-grid-4-tutorial-for-distributed-testing/

In the grid you install nodes (machines) that contain a fixed version of a browser, where you install a fixed version of the driver.

Maintaining your own grid is time consuming, probably more expensive than using something like BrowserStack which run the grid for you.

Asking the remote webdriver for a version of an OS/Browser will search the Grid nodes to find a suitable Node to run the test on. On the node the driver.exe will be started, etc...

desired_cap = {
 'os_version': '10',
 'resolution': '1920x1080',
 'browser': 'Chrome',
 'browser_version': '62.0',
 'os': 'Windows',
}
driver = webdriver.Remote(
    command_executor='https://your_grid/wd/hub',
    desired_capabilities=desired_cap)
driver.get("https://www.google.com")

This results that in your local code you do not need to manage the drivers anymore.