GLM-5.2 Tops Open Weights Models on Artificial Analysis Index
Original: GLM-5.2 is the new leading open weights model on Artificial Analysis
Why This Matters
Demonstrates competitive capability of open weights models versus proprietary systems in agentic reasoning tasks, affecting model selection for enterprise deployments.
Zhipu AI's GLM-5.2 achieved a score of 51 on the Artificial Analysis Intelligence Index v4.1, ranking as the leading open weights model ahead of MiniMax-M3 (44) and DeepSeek V4 Pro (44). The 744B parameter model shows 11-point improvement over GLM-5.1 while maintaining identical size and similar API pricing at $1.4/$4.4/$0.26 per million tokens.
Zhipu AI released GLM-5.2, which claims the top position among open weights models on the Artificial Analysis Intelligence Index v4.1 with a score of 51. The model maintains the same architecture as GLM-5.1 (744B total parameters with 40B active parameters) but demonstrates significant performance gains across multiple benchmarks. GLM-5.2 expanded its context window from 200K to 1M tokens. On scientific reasoning evaluations, the model showed particularly strong improvements: CritPt increased by 16 points to 21%, HLE gained 12 points to 40%, and SciCode improved by 7 points to 50%. The model also achieved 89% on GPQA Diamond. On GDPval-AA v2, a benchmark measuring real-world agentic performance, GLM-5.2 scored 1524, outperforming MiniMax-M3 (1418) and DeepSeek V4 Pro max (1328), and performing comparably to proprietary GPT-5.5 (xhigh reasoning, 1514). The model uses 43,000 output tokens per Intelligence Index task, with 37,000 devoted to reasoning—higher than GLM-5.1 (26k) and peers MiniMax-M3 (24k) and Kimi K2.6 (35k). API pricing remains aligned with GLM-5.1 at $1.4/$0.26/$4.4 per million input/cache hit/output tokens. GLM-5.2 is available through Zhipu AI's first-party API and multiple third-party providers including DeepInfra, Novita, Nebius, Parasail, Siliconflow, and Fireworks. The model operates under an MIT license.