What's the lowest phase noise OCXO you've actually put on a bench?

I've been working with precision timing components for longer than I'd like to admit, and I've tested a lot of oscillators over the years. Most high-performance OCXOs land somewhere around -155 to -160 dBc/Hz at 1kHz offset, which is already pretty impressive if you've ever had to debug a noisy clock.
But the best one I've personally worked with hit -170 dBc/Hz at 1kHz offset.
For context, that's another factor of 10 quieter than a -160 dBc/Hz part — the kind of performance you'd normally associate with a rubidium atomic reference, except it's in a normal DIP package you can solder right on your PCB, at a fraction of the cost and size. When we first measured it on the bench I had to double-check the setup because it felt like the numbers shouldn't be possible with a crystal.
I'm curious — what's the lowest phase noise OCXO you've actually used in a real design? What were you building, and how much phase noise did your system actually need?
Drop it in the comments, I'm always interested to see what other engineers are working with.