ByteDance Builds China's Largest AI Model

According to the Financial Times, ByteDance is training an AI model with up to ten trillion parameters, making it the largest in China. This is three times larger than Moonshot's Kimi K3, the previous largest Chinese model, and approaches Anthropic's Mythos 5, which industry estimates put at around eight trillion parameters, though Anthropic has not disclosed official numbers. Three insiders told the FT that the model is in pre-training, a phase that usually lasts three to six months. Parameters determine how much a model can store, but performance also depends on data quality and training methods. One source said ByteDance has refrained from distillation, or fine-tuning on other models' outputs, for over a year. Founder Zhang Yiming has urged the 2,000-strong Seed team to pursue world-leading model capabilities in the long term. Meanwhile, Elon Musk's xAI is reportedly training Grok variants with six and ten trillion parameters on its Colossus 2 cluster.