Zai GLM-5.3-Flash Model Confirmed as Ox Alpha Leader

Zai has confirmed its GLM-5.3-Flash model as the Ox Alpha model, achieving impressive performance metrics.

The model boasts 320 billion parameters and 18 billion active per token, all while maintaining a cost-effective $0.09 per task.

This achievement was made possible through aggressive memory optimization on China-made hardware.

DeepLearning.AI highlighted the LLM architecture details in its latest Batch report, underscoring gains in AI model performance and real-world deployment efficiency.