Taiwan's Foxconn has launched its first Large Language Model (LLM), FoxBrain, to enhance manufacturing and supply chain operations. Based on Meta's Llama 3.1, the AI model features 70 billion parameters and a 128k-token context window. It was trained in four weeks using 120 Nvidia H100 GPUs, scaled with Nvidia Quantum-2 InfiniBand networking, and employs adaptive reasoning reflection to improve efficiency.
Also Read: Tech Mahindra Launches AI Network Automation Model for Telcos With Nvidia and AWS
Foxconn Unveils FoxBrain
"The institute [Hon Hai Research Institute], backed by Hon Hai Technology Group (Foxconn), the world's largest electronics manufacturer and technological solutions provider, said the LLM – code-named FoxBrain – will be open-sourced and shared publicly in the future. It was originally designed for applications used in the Group's internal systems, covering functions such as data analysis, decision support, document collaboration, mathematics, reasoning and problem-solving, and code generation." Foxconn said on March 10, 2025.
"In recent months, the deepening of reasoning capabilities and the efficient use of GPUs have gradually become the mainstream development in the field of AI. Our FoxBrain model adopted a very efficient training strategy, focusing on optimising the training process rather than blindly accumulating computing power," said Yung-Hui Li, Director of the Artificial Intelligence Research Center at Hon Hai Research Institute. "Through carefully designed training methods and resource optimisation, we have successfully built a local AI model with powerful reasoning capabilities."
Also Read: IBM and L’Oreal to Build AI Model for Creation of Sustainable Cosmetics
Technical Capabilities
Foxconn claims that FoxBrain outperforms Llama-3-Taiwan-70B in math and logical reasoning but still lags behind DeepSeek's model.
"In test results, FoxBrain showed comprehensive improvements in mathematics compared to the base Meta Llama 3.1 model. It achieved significant progress in mathematical tests compared to Taiwan Llama, currently the best Traditional Chinese large model, and surpassed Meta's current models of the same class in mathematical reasoning ability. While there is still a slight gap with DeepSeek's distillation model, its performance is already very close to world-leading standards," Foxconn added.
The company stated that this large language model research demonstrates Taiwan's technological talent can compete with international counterparts in the AI field.
FoxBrain's Future
Foxconn also announced that, in the future, the Group will continue collaborating with technology partners to expand FoxBrain's applications, share its open-source information, and promote AI in manufacturing, supply chain management, and intelligent decision-making.
Also Read: Meta Unveils New AI Models and Tools to Drive Innovation
Foxconn and Nvidia
During model training, Foxconn said Nvidia provided support through the Taipei-1 Supercomputer and technical consultation, enabling Hon Hai Research Institute to complete the model pre-training with Nvidia NeMo. "FoxBrain will also become an important engine to drive the upgrade of Foxconn’s three major platforms: Smart Manufacturing. Smart EV. Smart City," the official release said.
The company will present its results at Nvidia GTC 2025 on March 20.
