Kimi K3 vs DeepSeek V4 Pro vs GLM-5.2: trillion-scale MoE models compared

Moonshot AI's Kimi K3, DeepSeek V4 Pro, and Zhipu AI's GLM-5.2 — all sparse MoE models with million-token context windows — top the open-weight leaderboard. The comparison covers benchmarks, licensing, and serving cost, with all three targeting long-horizon coding and agent workloads.