Two ways to compare. First is per pop. A mature cottage should net ~23 beakers per pop (7C -> 21 beakers, 1H -> 2 beakers).
A windmill can get you ~16 beakers per pop: 4C -> 12 beakers, 2P -> 4 beakers.
A bio-farm plus a scientist spec., plus bulbing with the GS that pop, should get you ~10-12 beakers per turn per pop on average (18 for spec., 2-6 for bulbing, split over two pop).
A workshop and building research should net 8 beakers per pop (4H -> 8 beakers).
A bio-farm plus a spy spec. is good for 6 beakers and 6 espionage per pop. The 6 EPs should be worth anywhere from 10-30 beakers in stolen techs depending on other game factors (like whether you got a holy city), making this worth 16-36 beakers per pop.
From this perspective, cottages look very good. Windmills can give you 75% of the research, spy specs. can actually do better if you've got all the right infrastructure, the right game situation, and good neighbors to steal from. Everything else is doing far worse - as in, barely half as good. And this is the right perspective if the limiting factor on your development is happy cap.
But the other way to look at it is per-tile. If you're limited not by happy cap, but by amount of land you can make use of, this is a better approach. Note that just because all your current cities are at happy cap doesn't mean you are limited by happy cap - you can settle new interior cities that overlap tiles. A late-game bio-farm + specialist economy will tend to have slightly denser city placement than a late-game cottage economy, because you need more cities to work all your tiles.
Per-tile, cottages are still worth 23 beakers.
Windmills are still worth 16 beakers.
Bio-farms plus scientist specs are worth 20-26 beakers; comparable to cottages.
A workshop is still 8 beakers.
Bio-farms plus spy specs. are worth 32-72 beakers.
Here cottages don't look so unassailable. In fact, bio-farms and spies are blowing everything else out of the water.
But now the analysis starts to break down into more nebulous factors. Settling more cities to work the same number of tiles plus some specs. is going to increase upkeep significantly. Relying on spy specs. means relying on your neighbors' tech choices, and might well piss off your neighbors as well. Cottages are very vulnerable to pillaging. Workshops give you a big boost to flexibility of production in addition to their research potential. Windmills have to be built on hills, where you might want mines instead, and require Environmentalism to hope to compete. Farms and specs. allows for whips to boost production. A whole host of early-game factors might push you into going for farm-and-spec. early, in which case a late-game transition to cottages is quite difficult. Farms give more pop, which gives you more votes in the UN. PHI leaders get more GP; ORG leaders can more easily pay the upkeep of late-game specs, CHA leaders can be somewhat less limited by happy cap late game. FIN leaders get stronger cottages. And the list goes on...
Bottom line, though: most of the time, the only thing that can keep up with mature cottages in research late-game is going to be spy specialists, and even that is a bit situational. But - more densely packing your cities can get scientist specs. to come close. Also, mature cottages are one of the weakest late-game options for production (bio-farms plus slavery, and particularly plus slavery and Kremlin, are much better; workshops, windmills, mines, lumbermills, and watermills are all obviously better).
