GLM-5.2 just released and the early numbers look pretty insane. 1M context window, open weights, MIT license, two reasoning effort modes, and it is already showing up near the top of coding arenas. I know every new model gets hyped for 24 hours, but this one actually looks worth testing because it i
