the problem of clock timing runs pretty deep - there are two major issues
1] midi clock needs to be very, very, stable. It is sent at too low a resolution (24 of these from one beat to the next). There’s the sending half and the receiving half of this problem - the device sending clock must have it implemented so no other function ever interferes with when the clock pulse (midi message) goes out. When the clock master device also plays audio, and allows you to edit things on a display, it could get messy. The receiving half is subject to timing issues as the message travels - the weakest link is usually the MIDI<>USB boundary, but also USB<>USB (!). Old USB interfaces/controllers/devices generally have worse timing performance.
2] then there’s the OP1 tape, on which one has a looped segment that they expect to loop in sync. If we’re doing this in similar ways, I go to a beat marker, loop in, then to the next, loop out, and loop-play the section. The problem is hard to see, but as the clock drifts, and this is where a DAW differs significantly, the OP1 keeps the loop length the same, i.e. the same number of samples. A DAW, in comparison, would adjust the loop points because they’re marked in beats and bars, not sample rate samples, so the loop length may change and all MIDI played would still sound in sync. With DAWs like Ableton that stretch audio, the audio also sounds in sync. In fact, if the timing changes are small, the adjustment of the loop boundaries would have you not notice the issue with non-stretched audio, but on a DAW the adjustment is continuous and always applies. On the OP1 there is no adjustment. It loops, then it’s off, so it loops when the loop ends, not when “it’s time to loop”.
Unless the OP1 starts dealing in tempo ticks, which I doubt it ever would (, but one can dream!), you’d need to move mountains to get it to play nice as sync slave. Using it as a master would be assuming that it itself doesn’t mess up its own timing, especially as it tries to smooth audio over the loop points. Something however tells me that it does.
So, contrary to all reason, monkey-sync by just keeping the same BPM actually ends up being more reliable.