Taiwanese electronics giant Foxconn has announced the development of its own large language model (LLM), FoxBrain, which has been trained in just four weeks. The LLM is designed to support internal operations, including data analysis, mathematics, reasoning, and code generation.
Initially developed for internal applications, FoxBrain is expected to enhance Foxconn’s efficiency in manufacturing and supply chain management. The company has indicated plans to open-source the model, allowing industry partners to leverage its capabilities for broader applications, reports WSJ.
Foxconn credited Nvidia for its role in FoxBrain’s development, with the chipmaker providing access to a Taiwan-based supercomputer and technical consulting. According to Yung-Hui Li, director of Foxconn’s AI research centre, the AI model was trained using 120 Nvidia H100 graphics processing units (GPUs), prioritising optimised training strategies rather than sheer computational power.
Foxconn is based on Meta’s Llama 3.1 framework and has been optimised for traditional Chinese, the primary written language in Taiwan and some overseas Chinese communities. The architecture has 70 billion parameters and features 128K tokens for its context window length, reports Analytics India.
Foxconn claims its model’s reasoning capabilities place it slightly behind some offerings from China’s DeepSeek but approaching world-class standards.
The company would reveal more information at Nvidia’s annual technology event in mid-March in a session talk titled ‘From Open Source to Frontier AI: Build, Customise, and Extend Foundational Models.’
“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. “Through carefully designed training methods and resource optimisation, we have successfully built a local AI model with powerful reasoning capabilities.”
In the News: Delhi Police arrests four in Rs 44.50 lakh digital arrest scam
