Curious as to why BL went they way they did with their filament changes - ie, every filament is cut before retraction, causing a lot of wastage. Why is this better then Prusa’s MMU3, for instance, where the entire filament is retracted?
Other then producing good filament tips for easy retraction/reloading, I can’t see the benefit to BL’s implementation? Even then, Prusa seem to hav ethis cracked with far, far less wastage - even though it’s not as ‘neat’ a solution as the AMS…
It really comes down to a design decision. The Prusa audience is far more technical than the Bambu audience, and Bambu’s design is geared for set-it-and-forget-it use, although the state of the art is not yet at the level of a desktop laser printer. Bambu has come the closest.
And on the topic of filament waste? Is under 10mm really worth it for savings? There is a point of diminishing returns vs simplicity of use. Mind you I’m not defending the approach just merely offering an explanation.
Doesn’t mmu also cut the filament? I have developed a brand of 3D printers with direct extruders and the most common fault was that cura decided to revert the reaction back to 6.5mm. Reacting hot filament goes well 99/100 times but that one time it doesn’t the hotend is clogged and the print is bricked. And 1/100 retractions is basically every print
It does have a blade, but only uses it to cut if it cannot retract the filament after 3 failed attempts.
I do love the BL way of woking to be fair - its very beginner friendly and just works, and the filament is much cheaper then Prusament too (refills even more so!) so I guess some wastage is fine, but it’s when the wastage gets to a point where it outweighs the actual model that it grates on me…
Currently building a MMU3 for my MK4s so will test that and see how it goes in order to compare… 
I may get to a point where I print actual models with the MMU3, and any nameplates/coasters, and other things with my A1 where its just the first few layers that need colour…
It’s only a guess but just pulling back without cutting can also leave thin strings in the filament path to contaminate other colors.
Just my opinion but it’s worth wasting an inch or so of filament to not have to worry about that. If cutting before pullback was optional I’d still do it.
The reason the Prusa mk4/mk4s with MMU3 produces much less filament waste (and does the swap faster) is the nozzle. Bambulab (and Anycubic on the Kobra 3 combo) use higher flow nozzles. On standard-flow nozzles you can get a nice filament tip on unloading using ramping. Prusa did a good job developing these.
If using nozzles with higher flow, the filament-tip is not as good. You can not load it reliably again. Hence why Prusa recommend swapping to a standard nozzle when adding the MMU3 to the MK4s, which uses a HF nozzle by default.
As a result it seems Bambu Lab went with speed and the HF nozzle, and the most reliable way to implement reliable tips was to cut the filament each time, which of courses produces all that waste. Its just a different implementation to the same process. Although it’s cleaner, and allows the use of a HF nozzle, it isn’t necessarily making the print job faster, since the BambuLab may print faster due to teh HF nozzle, but the filament changes take an age compared to Prusa, where a filament change is far quicker in comparison.
Of course the ideal solution would be a dual nozzle hot end. If you want multicolour it switches to the standard nozzle (fast filament change and low waste) - for single colour it switches to a HF nozzle.
The BL way of cutting and ejecting waste is OK if you only have a few filament swaps. Or maybe only print multicolour once in a while. This is how I use it - for coasters or name plates etc that only require the first 2 or 3 years to be multicolour, then switch to single colour for the remainder of the print. Whereas the Prusa way of ejecting a refined tip seems to be better for actual multicolour models that have many colour changes, which are fast and produce minimal waste.