Z.AI: GLM 4.6

Z.AI: GLM 4.6, released by z-ai on September 30, 2025, is a significant upgrade from GLM-4.5, featuring an expanded 200K token context window for complex agentic tasks. This model demonstrates exceptional reliability with a 97% success rate across benchmarks, indicating consistent operational stability. However, it tends to have longer response times, ranking in the 10th percentile for speed, and is positioned at premium pricing levels (12th percentile). Performance highlights include perfect accuracy in General Knowledge and Email Classification, with these benchmarks also noted for being the most accurate at their respective price points and speeds. GLM 4.6 excels in Coding (95.0% accuracy, 94th percentile) and Mathematics (96.0% accuracy, 98th percentile), aligning with its described superior coding performance and advanced reasoning capabilities. Its hallucination rate is low at 96.0% accuracy. A notable weakness is its Instruction Following, where it scored only 3.1% accuracy (22nd percentile), suggesting challenges with multi-layered or complex directives. Ethics performance is also relatively low at 82.0% (21st percentile).