DigiNews

Tech Watch by Johan Denoyer

← Back to articles

GLM 5.2 Performance Benchmarks

Quality: 8/10 Relevance: 9/10

Summary

GLM-5.2 (max) from Z AI is benchmarked across intelligence, speed, and cost in Artificial Analysis' open-weights class. The report ranks it among the top intelligence models with a score of 51 on the Artificial Analysis Intelligence Index, achieves 112.4 output tokens per second, and lists 753B total parameters with 40B active. It notes high cost per task relative to peers, MIT license, and availability via Hugging Face; it uses a 1M token context window and highlights both strengths and trade-offs.

🚀 Service construit par Johan Denoyer