What happened
Foxconn launched its first large language model (LLM) named "FoxBrain," designed initially to support its internal manufacturing and operational functions. The model was developed in about four weeks using 120 Nvidia H100 GPUs and is based on Metaâs Llama 3.1 architecture. Foxconn plans to open-source FoxBrain in the future to expand its applications beyond internal use.
Why it matters
- FoxBrain aims to improve manufacturing efficiency through AI-driven data analysis, decision support, and problem-solving.
- The model is optimized for Traditional Chinese, addressing a niche underserved by many existing AI language models.
- Open-sourcing FoxBrain could foster collaboration and innovation in AI applications across manufacturing and supply chain management.
Key facts
- 1. FoxBrain was developed in approximately four weeks using 120 Nvidia H100 GPUs.
- 2. The model is based on Metaâs Llama 3.1 architecture and excels in mathematics and logical reasoning.
- 3. Foxconn announced the launch on March 10, 2025, and plans to open-source the model without a specified release date.
