GLM-5.2
Zhipu AI class A live
Zhipu AI's ~753B MoE (40B active) open flagship; 1M-token context, reasoning + agentic coding, MIT weights. Leading open-weights model.
Layer 1 — factual axes
Layer 2 — output quality
51 / 100
Zhipu AI’s ~753B MoE (40B active) open flagship; 1M-token context, reasoning + agentic coding, MIT weights. Leading open-weights model.