The important ones, basically. It's not 1990 anymore. There have now been many predictions made for times now in the past based on the same climate models being used to predict disaster, and they are now almost uniformly wrong.
The Met Office in Britain is one of the clearest. Are they predicting a wet summer for Britain? Brits should buy suntan lotion. Are they predicting a hot, dry summer? Brits should plant something that likes wet, dark conditions. They are the source of one of the models in the IPCC report, and they're predictions for a year or two out are wrong, consistently.
And I find the idea that a climate model can't predict a year in advance, consistently, but can be used for 10 or 100 years in advance to be absolutely ludicrous. Utter mathematical illiteracy is required to think that is true of a chaotic system. It remains unclear to me whether climate researchers actually understand that "a chaotic system" is actually a mathematically meaningful statement and not "an excuse for why we were wrong, again". (No sarcasm, it truly remains unclear to me. Chaos theory is a real thing, and I'd really like to think that climate researchers would know a lot about it, but I've only seen faint traces of it, where I expect the terminology to dominate the discourse.)
Given the utter failure of these predictions on a year and a decade scale (as I've said before, had the climate scientists said "hey, there's going to be a 15-year plateau because of this and this which will mask the problem but then this is going to happen", I'd be fine with that), I see no reason to believe they'll work on a century scale.
At the moment, the most reasonable explanation is that the models used by the IPCC up to this point have been grossly wrong about their positive heating feedback claims, which if you dig, were never really that well substantiated by actual evidence, they have always been modeling choices. In the real science world, those claims are being walked back quite rapidly with a variety of peer reviewed studies providing both evidence of this and potential models for explanations; this has not penetrated to the political portion of the AGW crowd yet. This is probably because the science, to my eyes, is rapidly converging on "global warming is unlikely to be a problem in this century, and may even be a net benefit to humanity". I wouldn't be surprised this is the dominant science consensus in ten years.