GLM-4-AirX is the high-speed enhanced variant of GLM-4-Air from Zhipu AI. Built on MoE architecture, it retains comparable model capability with greatly accelerated inference. It supports function calling, code generation and Agent tasks, optimized for low-latency, high-concurrency API services.