← Retour au blog
tech 17 June 2026

GLM-5.2: The New Leading Open Weights Model on the Intelligence Index

Z ai's GLM-5.2 transforms the open weights model landscape with its impressive Intelligence Index score. Learn how this model outperforms its competitors.

Article inspired by the original source
GLM-5.2 is the new leading open weights model on Artificial Analysis ↗ artificialanalysis.ai

Introduction

In the ever-evolving world of artificial intelligence, each technological advancement can be a game-changer. This is exactly what GLM-5.2, the new open weights model from Z ai, is doing, having recently taken the lead in the Artificial Analysis Intelligence Index. With a score of 51, it significantly outpaces other open models and even some renowned proprietary models.

Why is GLM-5.2 a Leader?

GLM-5.2 stands out due to its balance between intelligence and cost per task. While other models like MiniMax-M3 and DeepSeek V4 Pro have respective scores of 44, GLM-5.2 advances with increased efficiency and significantly improved scientific reasoning capabilities. Its improvement is visible across most evaluations, notably in scientific reasoning, where it records significant gains.

Outstanding Performance

  • Scientific Reasoning: GLM-5.2 shows a 16-point gain on CritPt, reaching 21% in scientific reasoning.
  • HLE: It progresses by 12 points to reach 40%.
  • TerminalBench v2.1: Improvement of 16 points, reaching 78%.

These figures not only demonstrate GLM-5.2's lead over previous models but also its competitiveness against cutting-edge proprietary models.

Comparison with Other Models

GLM-5.2 uses 43,000 output tokens per Intelligence Index task, much more than other open models like MiniMax-M3 with 24,000 tokens. This indicates increased complexity and depth in task processing.

Cost Per Task

Although GLM-5.2's cost per task is slightly higher (~$0.46 per task) than some competitors, its position on the Pareto frontier between intelligence and cost shows that it offers the best value for its performance level.

Technical Advantages

GLM-5.2 features 744 billion total parameters, of which 40 billion are active and uses an enhanced context window of 1 million tokens. Its availability on various platforms like DeepInfra and Novita makes it accessible to a wide range of users.

Conclusion

Z ai's GLM-5.2 is not just an incremental improvement. It is a significant step forward in the realm of open weights models, offering increased intelligence and optimized cost per task. If you're looking to integrate a high-performing AI into your processes, GLM-5.2 is an essential choice.

Let's discuss your project in 15 minutes.

GLM-5.2 Artificial Intelligence Open Weights Models Intelligence Index AI Performance
Deepthix newsletter · 100% AI · every Monday 8am

An AI agent reads tech for you.

Our AI agent scans ~200 sources per week and ships the best articles to your inbox Monday 8am. Free. One click to unsubscribe.

Visit the newsletter page →

Want to automate your operations?

Let's talk about your project in 15 minutes.

Book a call