Taizhiyun Technology's independently developed Formosa Model (FFM) is the world's first enterprise-grade optimized Traditional Chinese language model, aiming to deepen the development of sovereign AI, strengthen data security, and promote industrial upgrading. FFM is trained based on mainstream international open-source large-scale models. In just two years, it has continuously iterated and improved multiple large-scale models that retain the capabilities of the original models while understanding Traditional Chinese cultural semantics and industry knowledge. The latest Llama3.2-FFM visual multimodal Traditional Chinese model significantly improves the accuracy of Traditional Chinese image recognition, understanding, and question answering.
FFM has passed the AIEC testing and certification of the Digital Development Department, ensuring security and localized application, and supporting flexible cloud and ground deployment and commercial licensing. Combined with a one-stop AI model optimization and inference service and platform tool (AFS), it helps enterprises to adopt and transform AI, accelerates the service of the upstream and downstream vertical AI industry chain, and promotes the implementation of AI innovation with high efficiency, low threshold and large scale.
It has been successfully applied in various industries, including smart cities (5G water recycling), manufacturing (optoelectronic processes/metal fasteners), finance (risk control and review), healthcare (clinical/ECG analysis), retail (intelligent customer service), ESG (carbon data), and terminals (mobile phones/laptops). It supports the development of national sovereign AI and the industrialization of AI, driving Taiwan's global AI competitiveness.