Foxconn is also now entering the arena of producing a public language model (LLM) that will be the basis for its own artificial intelligence (AI). Today they announced the FoxBrain LLM, which was trained for four weeks using 120 NVIDIA H100 GPUs. This LLM was developed to improve factory management and their supply chain.
As one of the world's largest device manufacturers, Foxconn is responsible for manufacturing several Apple products such as the iPhone and MacBook. According to them, FoxBrain is based on Llama 3.1 with 70 billion parameters. It is the first LLM developed in Taiwan that can perform reasoning and is optimized to understand traditional Mandarin and Taiwanese Mandarin.
It was developed to work with several applications currently used by Foxconn. Among its capabilities are data analysis, mathematical calculations, code generation, reasoning, document understanding and helping employees make decisions. In terms of performance, it is still behind DeepSeek but according to Foxconn it is still comparable to several other international AI models.