The models have underpredicted the temperature rise throughout my entire living memory. Even the "pessimistic" scenarios of high-temperature-rise have been blasted right through on a yearly basis. The model is drastically, dramatically understating the natural feedback loops involved and the amount of temperature rise that's going to happen.
Feedback loops like methane release are likely also hugely understated, as another practical example.
A large amount of this is due to political manipulation and pressure. Researchers are afraid of the way the issue has been politicized, and generally don't want to go out on a limb for fear of personal reputational damage or consequences to the field of study as a whole. Even just 15 years ago, in the 2000s, conservative-dominated media was begging for someone to step over the line so they could cancel them Dixie Chicks style and pronounce the entire field of study as discredited.
People spend way too much time on "but what if the models are overstating the problem" and they don't, actually they aren't even centered on the actual rate of increase, we pre-manipulated the model by building in all these stabilizing assumptions and even the pessimistic scenarios are actually quite optimistic. And then people apply additional optimistic "cheater factors" on top of that, and we still can't get to a situation in which the climate is not drastically deteriorated in a hundred years.
As I said, we have blasted through the most pessimistic assumptions on a yearly basis, the models are very clearly not aggressive enough about the temperature rises that are going to happen this century.