I think this has all come up before. but IMO the biggest difference between Serge and Eurorack (even with R*S modules) is the overall ergonomics and workflow.
Yes, in Euro, as @909one says, there are myriads of options for sequencers, controllers, etc etc. This promotes a way of working and thinking about the role of a module. Even though lots of feedback patching, multi-purpose modules, and stacking (with the right cables) are possible, it is not what is incentivized by the design of the system and modules. And many many Euro modules are not really multi-purpose or patch-programmable (in the Serge sense), and definitely don’t lend themselves to that way of thinking.
In a full Serge banana system there are fewer specific “control” options provided as modules, but the workflow and ergonomics incentivize and encourage a different way of thinking… much more about using slopes and pulses as controls than about dedicated sequencers for example. More about feedback and self-patching, making the low level modules into what you need for the patch rather than getting specific modules for different functions.
And while some Serge circuits definitely have a unique sound quality, my view is that a lot of what sounds like “Serge” is more because of the workflow and ergonomics then the specific slopes of an oscillator. And while you can achieve similar workflows with Eurorack, it is definitely not the overarching design philosophy of the system and requires some workarounds to achieve what the banana systems are built for. The same is true in reverse. If you want lots of digital modules, dedicated sequencers and controllers, and really specific things (like the referenced Marbles), then Eurorack is more suited to those types of options.