The Graph Protocol today launched and open-sourced GLM 5.2, designed for long-range mission capability

June 17th: Zhizhi (02513.HK) announced today the launch and open-sourcing of its GLM-5.2 model. On the frontend development assessment platform CodeArena, with participation from millions of users worldwide, GLM-5.2 has achieved top performance among all globally available models. Designed specifically for long-range task capabilities, GLM-5.2 comes with the following new features: - 1M context window, providing stable support for long-range tasks; - Enhanced sensory experience and improved practical coding abilities; - Optimized infrastructure, enabling smooth operation on domestic computing power platforms from day one; - MIT open-source license, free from geographic restrictions, to promote cross-border technological equality. The 1M context window forms the core of GLM-5.2’s long-range performance. Multiple long-range task benchmarks show GLM-5.2 ranks between Claude Opus 4.7 and 4.8, making it the highest-performing open-source model globally. GLM-5.2’s online inference is supported by multiple domestic computing platforms. It completed inference adaptation with domestic systems including Huawei Ascend, Flathead Brother, Moore Threads, Cambricon, Kunlun Core, Muxi, Haiguang, and Biren on day one, now running stably on domestic chip clusters with high throughput, low latency, and strong concurrency. The upcoming launch of Huawei’s Ascend 950 Super Node in the second half of the year is expected to further boost GLM-5.2’s computing foundation. --------------------------------- Click the original link below to join the Motion-Driven · Feishu AI News Channel for 24/7 real-time monitoring of global AI hot topics and news.