actually, unless you bend the legs of the stage with heat (like on bimetal-strips), i would assume, that any metal version of the center-stage would fatigue MUCH quicker, than any plastic-printed one.
Lets leave SLM-printing aside for a moment, and just assume machined or cast steel parts. during the process of manufacturing the metal cools down unevenly around something called cristalization points.
The only thing that matters is, that the metal has some inherent defects in it’s structure on various levels of magnification. the level we are interested here is the crystalline grid that forms all metals on a microscopic layer. defects spread out somewhat uniformly thoughout the whole metal part gives it some interesting properties. When physically deforming the metal (in a cold state) beyond it’s flexible limit, those defects clash together and inter-twine and form a much stronger bond at the affected area. however, because that bent section is now much stiffer, it is basically impossible to deform it back to original. the most quoted example for this are paperclips. straigthen them out, and try to bend them back in shape, there will always be additional kinks in the metal. Now imagine the flexure-stage to be bent back and forth alot. It would fatigue very quickly. To “reflow” the metal back to it’s more defective and more malleable state, you do something called tempering (in case of steel: heating it to a few hundred degrees and let it cool down reeeealy slowly).
So obviously Plastic deformation as described above isn’t useful here (unless induced by heat), and we want to stay in the limits of flexible deformation. I don’t have the exact numbers in mind, but I’d assume for most metals the flexible deformation range is likely much less than e.G. PLA. Most likely you’d have to get the bendy-parts VERY thin for this to work, and i suppose the rather bumpy surface of SLM-printed parts doesn’t do us a favor either here. if you don’t nail the perfect thickness 100%, the motors will not be able to bend the legs anyways. the sweetspot is most likely a very narrow range of thicknesses. all bending sections should ideally be machined instead. Another problem (on long, very thin metal strips that translate motion with a rather big ratio) would be, that you will have an excessive amount of drift induced by temperature changes. You really cannot produce “living hinges” like on a plastic lunchbox in metal, at least not within reasonable assumptions.
With much softer materials this will most likely work better, but gold, silver, lead or even soldering-tin are all wildly unpractical, expensive or harmful. copper might also work though.
However, using bimetal strips are still wildly impractical, but it would be a very fun idea nontheless, and it might even work reliably good, but you have to alter the construction quite alot, otherwise your o-rings would just melt…
(P.S.: I learned alot of metallurgy in my training way back in the day, however, i am not a material scientist, and have no specific numbers to back up my claims, but from 15+ years of experience working with metals daily, I’d assert that the above claims are somewhat grounded in reality. in the end openflexure is designed to be printed in plastic. if you want to build this in metal instead, much different design-considerations apply, even if you SLM-print it instead of just machining everything…
P.P.S.: as I speak german as my first language, and not english, some specific technical terms may by slightly off-spec 
)
