I think we’ll soon see exactly how good “reasoning” models actually are, as the next generation of problem-havers uncover things that were never discussed on SO or elsewhere in-depth.