As the lead faculty member at my college for anti-online-cheating efforts, I think the OP's initial plan is too lenient, and secondarily too subjective.
OP's initial plan:
My plan is to send an email to the class giving the cheaters a chance
to come forward for a reduced penalty (scale down the grade based on
the severity of the cheating). Otherwise, they will be reported to the
OP's motivation for this as per a comment:
... because this is the first offense (it is the first exam though),
so I don't think jumping to reporting them is a good action. I may be
In line with other answers, I do think the OP should follow the standard institutional policy for academic integrity cases as soon as possible. In addition to the prior reasons, I would add this:
If these are university students, then it seems to me overwhelmingly likely that what's happened is a reflection of prior habits they've been following for... maybe 12+ years now? I'd say at this point it's naive to think this is truly "the first offense". What if these students are cheating at work in every single one of their college courses, and then pleading "first offense" or "didn't know" (very common, and should be disregarded as utterly unbelievable), and so are given this allowance continually throughout their program sequence?
My broad guess is that they've probably been given many "first offense" allowances over time, they've been either unpersuasive (or worse: evidence there is never any real penalty), and as a university instructor who cares about academic integrity (you've already spent the time to investigate this!), it's time to apply the putative penalty, so as to get the actual message across.
Moreover, as others have stated, the central Academic Integrity Officer is likely to maintain university-wide records, and decide or recommend increasing penalties for students who have had prior reports filed. (At my school, the available reports actually span 25 different campuses in our university system, in which transfers are common.) Having every instructor silo their own "first offense" process short-circuits that mechanism.
For these reasons, under the assumption that university students are expected to be previously aware of the rules (perhaps by reading syllabus information or other student materials, etc.), and also fairly long prior academic experience, I recommend assessing roughly the harshest penalty for cheating on tests possible. In my case, the default is a zero in any such case. I've found that assigning this as a "pending" null grade to the assignment makes students much more prompt about responding to cheating investigation inquiries (which otherwise go unreplied or "ghosted" in many cases). Perhaps more importantly, the OP's initial "scale down the grade based on the severity of the cheating" idea sounds vague and likely to result in bias or irregularities if they don't have a specific expected level of sanction decided in advance.
Also, I would generally avoid sending any course-wide messages out about the situation. That's because: (a) it's probably irrelevant, and possibly confusing, to the majority of students, (b) it fills the course mental space with negative chaff, and (c) it's probably a sign of instructor laziness, in that they couldn't bother to send a message to the specific students under investigation, which is the appropriate thing to do.