TL;DR
Zhipu's GLM-5.2 scores 51 on Artificial Analysis Intelligence Index, leading all open weights models with 11-point improvement over GLM-5.1 while maintaining same parameter count.
Key Points
- GLM-5.2 scores 51 on Intelligence Index v4.1, ahead of MiniMax-M3 (44) and DeepSeek V4 Pro (44)
- Same 744B total / 40B active parameters as GLM-5.1 but gains 11 points, particularly in scientific reasoning (+16 on CritPt, +12 on HLE)
- Scores 1524 on GDPval-AA v2 (agentic real-world tasks), competitive with GPT-5.5 xhigh (1514), outperforming proprietary peers
- Context window expanded to 1M tokens from 200K; pricing $1.4/$0.26/$4.4 per 1M input/cache/output tokens; available on 8+ third-party providers
Why It Matters
For engineers evaluating open weights models, GLM-5.2 demonstrates that parameter efficiency and reasoning capability can advance significantly without scaling model size—critical for cost-conscious agentic AI deployments. The 1M context window and strong GDPval-AA performance make it competitive with proprietary models for real-world knowledge work tasks.
Source: artificialanalysis.ai