Yeah, ChatGPT is what they call a "trustworthy source". Not.
I'm inclined to say that it is exclusive. My first thought was like yours: it depends on when the condition is evaluated. But then it occurred to me that msdb.dbo.sysjobschedules has the columns next_run_date and next_run_time. I set up a job to run every minute with 23:59:00 as the end time. At 23:58 I checked that the job had executed at that time, and then I check the table. The was not there. So already at the execution at 23:58, Agent decided not to run it at 23:59.