AI BriefingGLM-5.2 Takes the Top Spot Among Open-Weight Models on Artificial Analysis
·2026.06.18 09:51
Key point
Z ai's GLM-5.2 recorded the top performance score among open-weight models on Artificial Analysis metrics.
Details
Z ai's GLM-5.2 scored 51 points on Artificial Analysis's Intelligence Index v4.1, surpassing MiniMax-M3 (44 points) and DeepSeek V4 Pro (44 points) to rank #1 among open-weight models.
Key features are as follows:
- Performance improvements: The model shows significant gains over GLM-5.1 across most evaluation metrics, including scientific reasoning (CritPt +16, HLE +12) and terminal benchmarks (TerminalBench v2.1 +16).
- Agentic performance: It scored 1524 points on the GDPval-AA v2 metric, demonstrating practical agentic task performance on par with GPT-5.5 (xhigh reasoning).
- Efficiency: The model sits on the Pareto frontier in the Intelligence vs Cost per Task chart, offering the lowest cost per task ($0.46) at that level of intelligence.
- Model specs: It has a total of 744B parameters (40B active), with the context window expanded from the previous 200K to 1M tokens.
- License and accessibility: Released under the MIT License, it is available not only through the Z ai API but also via various third-party providers such as DeepInfra and Siliconflow.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
Artificial Analysis Releases Small LLM Inference Benchmark for Mobile DevicesQwen3.8 27B Scores 52 on Artificial Analysis
Qwen3.8 Max Ranks First Overall on Agentic Index
Taste-Bench: Frontier LLMs Achieve 59.7% Accuracy on Decision Forking Tasks