I would consider them very similar, and certainly not the huge difference that KeePassXC has suggested. The problem with the zxcvbn algorithm is that it assumes anyone attacking the password knows the structure of the password. So in this case, the entropy calculations are based on the fact that an attacker knows the password is 6 dictionary words, with or without spaces respectively.
If someone was to attack those two passwords using a plain brute force attack without knowing the structure of the password, then the difference in entropy between the two passwords is 296 bits v 263 bits, with the 33 bit difference being explained by there being 5 more characters (with 95 printable ASCII characters in total) in the first password.
If you assume an attacker knows you are only using lower case letters and spaces, then that difference becomes 214 bits v 190 bits.
If someone was to attack those two passwords using a dictionary attack, then technically they are the same amount of entropy, somewhere in the region of 100 bits (assuming a 100,000 word dictionary), and which one was more secure would depend on whether your attacker was including spaces in their dictionary attack or not. I'm not sure, but I would assume most dictionary attacks do not include spaces, which would make the one with spaces more secure.
In some cases though, spaces would add a significant amount of entropy. For example, the passphrase:
demand fellow ship rail road happen
Is 6 dictionary words, whereas the passphrase:
demandfellowshiprailroadhappen
Is only 4, since fellow-ship and rail-road have been combined in to one dictionary word.
So, considering all of the above, I would add spaces.