Z.ai Launches Open-Weight GLM-5.3-Flash Multimodal Model

Z.ai’s GLM-5.3-Flash is a natively multimodal large language model aimed at combining frontier-scale capacity with lower inference costs. Its mixture-of-experts architecture activates only a fraction of the model for each request, improving computational efficiency, while a 1 million-token context window is designed for workloads such as large code repositories, lengthy documents and multimodal data. The release adds to intensifying competition in open-weight AI models.

Z.ai has formally released GLM-5.3-Flash with 320 billion total parameters and 18 billion active parameters, described as a 320B-A18B MoE configuration. The company is making the model weights available under the permissive MIT license, allowing commercial use and modification. Z.ai is positioning the model as a lower-cost option and says its coding performance approaches that of leading models in benchmark tests, though the supplied reports did not specify benchmark scores or a release date.