Zhipu identifies Ox Alpha as GLM-5.3-Flash and releases model weights
Zhipu has officially released GLM-5.3-Flash, confirming it as the model previously known anonymously as Ox Alpha. This new model is the first natively multimodal offering in the GLM-5 series, supporting various input types including text, images, and video. GLM-5.3-Flash features 320 billion total parameters and 18 billion active parameters. Its model weights are now publicly available through Hugging Face, following anonymous evaluations on platforms like OpenRouter and OpenCode.
Zhipu's release of GLM-5.3-Flash, with its 320 billion total parameters and 18 billion active parameters, marks a significant step for China's AI development. By making the model weights available on Hugging Face, Zhipu is directly challenging the established open-source AI landscape, particularly in Asia, where access to advanced multimodal models can accelerate local innovation. This move by Zhipu suggests a strategic push to democratize access to powerful AI tools within the region. The model's native multimodal capabilities, supporting diverse inputs from text to video, could drive new applications in sectors like media, education, and e-commerce across Asian markets. The key thing to watch will be the adoption rate and the quality of derivative applications built on GLM-5.3-Flash compared to Western alternatives.
Share this article
Related reading
6 stories
No need to destroy books in Singapore to legally use them for training AI, say experts
Zhipu's open-source model release raises questions about data use, a topic we explored with Singaporean experts.
China and US need to join hands against another powerful adversary: AI
Microsoft AI beats doctors in simulated diagnosis test - Fox News
Oracle widens its AI cloud menu with Google TPUs, custom chips, and Gemini data links
I made my coding agent explain its own pull requests
