One Nexus Releases Quantized GLM-5.3 and Flash Models

Hieu Pham announced on X that One Nexus has quantized the GLM-5.3 model and its Flash version to MXFP4 format. The resulting models are hosted on Hugging Face under the OneNexus organization. One Nexus operates as a neocloud in Vietnam providing AMD technology to users in Southeast Asia. Pham, who has worked at OpenAI, xAI, Google Brain, and Stanford, encouraged feedback on the quantized versions. The Hugging Face model cards highlight the goal of democratizing artificial intelligence via open source efforts.