That second sentence doesn’t imply the first one. The Samsung could have a lower false positive rate (letting the registered fingerprint unlock the phone more often) at the cost of a higher false negative rate (letting unregistered fingerprints unlock the phone more often)
In the limit, if it rolled a ten sided die for any press of its button, and unlocked the phone unless it rolled a one, that would give users the impression that it recognizes their fingerprint 90% of the time.
(Of course, “doesn’t imply” doesn’t mean the first sentence isn’t true, either. It may be genuinely better, or Samsung might have chosen a different cut-off, sacrificing security for a gain in usability. I wouldn’t know. Do any phone reviews test these things at all?)
The second gen was noticeably faster, I don’t know if it was more reliable.