Given that you've carved out the "realistic" exception for yourself I now presume you will keep shifting goalposts. But I digress.

System costs have been the central argument against renewables ever since they became competitive on LCOE. Which is why energy-system researchers have spent years developing, validating, and publishing models that account for hourly balancing, storage, transmission, flexibility, and sector coupling.

From Denmark to Australia system cost studies find renewable energy systems vastly cheaper than if involving new nuclear. Even when sensitivity testing very low nuclear costs.

The problem with nuclear is that it still needs the same firming to match an actual grid load. It still needs storage and gas turbines.

A smaller quantity than for renewable grids sure. But given how expensive new nuclear is it doesn't matter in the end.

Here's the Danish study:

https://www.sciencedirect.com/science/article/pii/S036054422...

And Australian:

https://www.csiro.au/en/research/technology-space/energy/Ele...

Please tell me how they aren't "realistic".

Your final comment conflates dispatchability with operational flexibility.

Nuclear output is controllable, and French reactors load follow. Although using a fleet controlling which reactor takes which day, and managing their fueling cycle so they can do it.

Conventional nuclear plants are economically optimized for high utilization, cannot start from shutdown on short notice, and don't eliminate the need for capacity that can respond to sudden shortages, demand peaks, and unexpected outages.

There is a significant difference between supplying dependable energy throughout the year and economically covering the relatively few hours when residual demand becomes exceptionally high.