Foxconn unveils 'Foxbrain' AI model aiming for world-class performance in Taiwan

Taiwan's Foxconn announced on the 10th that it has created its first artificial intelligence (AI) model, "Foxbrain." Foxconn explained that it successfully completed the AI model training in four weeks with the help of supercomputers and technical consulting from the American company NVIDIA.

According to the Wall Street Journal (WSJ), Foxbrain is a large language model (LLM) with reasoning capabilities, enabling data analysis, mathematics, code generation, and more. The new model utilized 100 NVIDIA H100 graphics processing units.

Foxconn plans to open source Foxbrain for collaboration with industry partners. Li Yonghui, Director of Foxconn's AI research center, noted that "we did not simply invest computing power into this model; we prioritized an optimized training strategy."

On the same day, Foxconn revealed only some specifications of Foxbrain and announced plans to share additional information at NVIDIA's annual technology event scheduled for mid-month. Foxbrain is reported to be based on Meta's LLaMA 3.1 architecture and is equipped with advanced reasoning capabilities. It was designed for traditional Chinese characters used in Taiwan and some Chinese communities.

Foxconn stated that the model's performance is approaching world-class levels, including some models from the Chinese AI startup DeepSeek.

Foxconn, well-known as the manufacturer of Apple iPhones, is facing difficulties in its electronics manufacturing business due to declining revenue and is pursuing diversification into areas such as AI and electric vehicles.