Follow the requirements of the journal, if they are firm. But let me argue why you should include all the authors, if possible, in many circumstances, even if there are three or four (or more). In particular, I come from a mathematics background, but this applies to many fields.
When you cite a paper as "X et al.,", the other author names are invisible. If the paper is good, and gets cited often in this way, people may begin to know it as the "X et al." paper - thus obscuring the contributions of the other authors. Thus X gets, in effect, sole credit in the text, and the other authors are relegated to the references section. The same holds at presentations.
Reputation is particularly important for many authors - especially younger ones, but even well established ones. It helps build their reputation for quality work, which in turn is related to jobs, grants, editorships, etc. The authors whose names are obscured may miss out on recognition that they actually deserve, solely so that an author can save a few characters in an electronic document.
This is particularly relevant in fields where authorships are alphabetical by default, such as mathematics. In this case, the first author only had the luck to have a name that comes earlier in the alphabet.
There have been studies where the effect of having an early-in-the-alphabet name have been investigated. Two of them are:
The essay "Et al." is unethical by Noah Snyder was influential in my thinking about this issue.