Post #2583210

idk I just don't get why you are making the argument that GLM 5.1 inference is profitable for providers at all here. The issue is something else. If you are seriously implying that GLM 5.1 is good enough that people would use it over $100 2028 Claude, then I guess we disagree about that. It's as if I said "training models could be an issue to do anonymously if Nvidia puts phone # verification in their GPUs and you have to buy a second one" and you replied with "a 2023 chromebook will be cheap forever" It's besides the point.