On August 26, Lingxi Professional Edition, an AI-driven office assistant developed by Kingsoft Office, made a significant announcement: the integration of the Zhipu GLM-5.3-Flash model. This model had previously undergone anonymous testing under the codename Ox-Alpha (referred to as the "Niulai Large Model" within the Chinese tech community) on the OpenRouter and OpenCode platforms. During this testing phase, it swiftly garnered widespread acclaim and set unprecedented records for invocation volumes on both platforms. GLM-5.3-Flash stands out as the inaugural native multimodal model within the GLM-5 series, boasting a cutting-edge architecture. Its performance is on par with leading overseas models, yet it comes at a fraction of the cost—just 1/40th of the price of comparable alternatives. Furthermore, its entire computational infrastructure is powered by 100,000 domestically manufactured chips, a testament to the capability of homegrown chips to efficiently fulfill the inference requirements of state-of-the-art models.