In a statement on Monday, Foxconn, Taiwan’s largest contract electronics manufacturer, announced the launch of its first large language model named “FoxBrain.” The model, trained using 120 Nvidia H100 GPUs over a period of four weeks, is intended to enhance manufacturing and supply chain management operations. Built on Meta’s Llama 3.1 architecture, FoxBrain is Taiwan’s inaugural large language model with reasoning capabilities tailored for traditional Chinese and Taiwanese language styles. Although Foxconn acknowledged a slight performance gap against China’s DeepSeek’s distillation model, it stated that FoxBrain’s overall performance is nearing international standards. Initially developed for internal use, FoxBrain encompasses functions such as data analysis, decision support, document collaboration, mathematics, reasoning, problem-solving, and code generation. Foxconn plans to collaborate with technology partners to expand the model’s applications, share open-source information, and advocate for the integration of AI in manufacturing, supply chain management, and intelligent decision-making. Nvidia provided support through its supercomputer Taipei-1 based in Kaohsiung, Taiwan, offering technical guidance during FoxBrain’s training. Further details about the model will be revealed during Nvidia’s GTC developer conference in mid-March.
