GLM 5.3 Flash Is INSANE — 320B MoE + Multimodal AI (= Ox Alpha)

Description

GLM-5.3-Flash is here, and it’s built for more intelligence with less compute. With 320B total parameters, just 18B active, hybrid attention, mHC, and 30T-token multimodal training, this model could be a serious AI contender. ============================================= Fo