AnalysisAI ModelsJune 19, 2026

MiniMax M3 vs GLM 5.2: coding benchmark

In a 60-task autonomous coding benchmark, GLM 5.2 achieved 92% full-pass (0.976 mean score) vs MiniMax M3's 84% (0.961). M3 was cheaper and faster: $6.67 vs $18.47 and 45s vs 80s average per run.

1 source