1. The memory difference in MODX vs. Montage affects how much memory user samples can consume. Everything else does not take up any significant space. The sample memory is shared between Libraries and User memory. The sum of all custom areas (Library plus User) sample memory is 1.0GB for MODX vs. Montage's 1.75GB.

2. File compatibility is fully documented here: https://usa.yamaha.com/products/music_production/synthesizers/modx/compatibility.html
MODX accepts Montage X7U (user) files, X7L (library) files. MODX is not compatible with Montage X7A (backup) files.

3. Size limit of X7U or X7L is not as important as the amount of sample memory the file will consume. Although technically feasible, it would be unwise for any single library to consume more than 1.0GB. The reason I say this is that, even on Montage with 1.75GB, if the Libraries consumed 1.0GB or more - you could not load two libraries. There are 7 library slots. Ideally, libraries would consume a maximum of 215MB each so that each area (Library 1-8) could have equal share of the total sample memory. However, in practice many of the larger commercial libraries are around 500MB - so you can load two at a time on MODX and three on Montage. Note that some libraries do not use samples at all - so some libraries may use zero sample memory.

4. Yes. Any with more than 4 PARTs to not support SSS. In general, there are single-PART versions of Performances which take up more than 4 PARTs so you can elect to use the slimmed down PART instead to conserve resources. This is speculation - but I fully expect that MODX will have firmware upgrades which take the 4-PART limitation in mind and further 'slim down" Performances so there is a wider variety. Something like "Seattle Sections 2" which perhaps uses a different approach to achieve the same goals within fewer PARTs (for example). Soundmondo will (eventually) also be a place where users can trade their customized versions which support SSS.

5. The FAQ https://usa.yamaha.com/products/music_production/synthesizers/modx/faq.html has, as its first item, the differences between MODX and Montage:

Sounds : Maximum polyphony, Part of SSS(Seamless Sound Switching), Flash capacity, Output jacks, MONTAGE has pure analog circuit

Control : Keyboard action, Numbers of Faders / Knobs / SCENE switchs [SIC], Aftertouch, Ribbon controller

Others : Weight, Dimensions, number of USB Audio connectivity

No aftertouch and no ribbon may or may not be important to you depending on how they are used on Montage.

Faders and scenes have buttons to access the same number as Montage - so you technically do not lose anything. The layout and gestures are just different. Assignable knobs fit in this category - but when using knobs to control parameter offsets (like "TONE", "MS/ARP", etc) - there are fewer parameters you can adjust using these knobs in single-function mode.

"Output Jacks" primarily is talking about lack of Assignable Output in MODX vs. Montage. This may not be important to you. There are some advanced routing options in Montage which has you route the Assignable output back around to the A/D input using a patch cable. This cannot be done in MODX. However, I would not consider this "normal" or "mainstream" use. There are more mainstream uses for Assign L&R outputs. MODX would probably add a mixer on Main L&R if you needed more analog outputs and also not worry about if effects are applied or not in that mix.

MODX supports 8 USB (digital) outputs while Montage supports 30.

Maximum polyphony is only different for FM-X sounds. MODX has half the polyphony as Montage for FM-X (64 vs. 128). MODX has the same polyphony for sampled sounds (128, same as Montage).

Damn, no aftertouch. That kills it for me 🙁

Such a shame, I was hoping for a lightweight gigging instrument. I suppose something had to give at that price.

MODX responds to aftertouch from an external controller. MODX's own local keyboard does not generate any aftertouch messages.