Length
Use when: Use 16 or more when the service allows it. NIST's current single-factor minimum is 15 characters; this control starts at 8 because some fields are still shorter, including passwords used only with a second factor.
Trade-offs: The upper bound grows linearly with length. A short pool, as in the 8-digit example, stays weak even though the form accepts it.
Character sets
Use when: Match the characters the destination accepts. Keep symbols on when they are allowed. The symbol set is !@#$%^&*()-_=+[]{};:,.?/ (24 characters).
Trade-offs: A set the site rejects wastes a generation. Uncheck it and generate again. The checkboxes do not make a human-chosen password stronger by themselves.
Avoid ambiguous characters
Use when: Someone may have to read or retype the password. The removed characters are 0, O, I, l, and 1.
Trade-offs: The full filtered pool shrinks from the unfiltered mix to 81 characters. Digits alone shrink from 10 to 8. That lowers the upper bound.
Uniqueness, storage, and the clipboard
Use when: You need a new value for one account. Store it in a password manager. Kitly does not save it, and a lost value cannot be recovered here.
Trade-offs: Each click draws again from Web Crypto. The tool does not check breach lists or your other accounts. Copy uses the clipboard, which the system or another app may keep. This page does not clear the clipboard afterward.
What the time figure leaves out
Use when: You want to compare two settings on this page. Read it as an illustration under the stated assumptions.
Trade-offs: It assumes uniform random choices and exactly 10^12 offline guesses per second. Phishing, reused passwords, and a breach of the destination are outside that picture. A high bit number does not prevent them.