June 18th.ZhipuYesterday's announcement on the lineOpen Source GLM-5.2, new model focused on Coding and a long-range mission。
GLM-5.2 has an automated research, performance optimization, complex system engineering and deep debugging, etc. Coding Agent scenario training
In the FrontierSWE long-range mission assessment, GLM-5.2 Max 10 Hrs scored 72.6%, down from 74.4% of Claude Opus 4.8 Max 20 Hrs, up from 63.0% of Claude Opus 4.7 Max 10 Hrs。
The model also introduces effort level thought-level control, which can switch between capacity, speed and cost. On the other hand, GLM-5.2 is known as the Day 0 fit for China's power platform, which is based on Chinese power, Hiranko, Moor, Cold Wu, Quinchurch, mussel, sea light, walls, etc。
💻 GitHub: github.com/zai-org/GLM-5
Hugging Face: Huggingface.co/zai-org/GLM-5.2
👾ModelScope: modelscope.cn/models/ZhipuAI/GLM-5.2