Bard told me that the number of Bitcoin addresses generated by a mnemonic phrase according to the BIP-39 standard is 2^126, while GPT said it is 2^32.
I don't know why people continue to ask these AI models anything technical, especially when it comes to bitcoin. They are consistently wrong and can easily be prompted in to providing any answer.
Some people say that the number of Bitcoin addresses generated by a mnemonic phrase is infinite, but I disagree.
It can't be infinite since there are a maximum possible number of bitcoin addresses.
then one mnemonic phrase would contain 2^256 addresses, which is clearly false.
If we are considering only one type of bitcoin address (such as P2WPKH addresses), then there are "only" 2
160 possible addresses, since each address encodes the output of a RIPEMD-160 hash. But it is correct to say that every possible P2WPKH address could be generated from a single seed phrase, if you scanned enough derivation paths. My answer that Zaguru12 has quoted above explains why. Doing this is obviously impossible, though, and is akin to simply brute forcing all possible private keys.