LongCat-2.0: China's Trillion-Param Model Trained on 50K Homegrown Chips Rivals Gemini
Meituan's LongCat-2.0 claims to be the first trillion-parameter model trained end-to-end on a 50,000-chip domestic compute cluster, matching Google's Gemini 3.1 pro. This achievement demonstrates that Chinese AI chips can now support large-scale training, intensifying the global AI race and loosening Nvidia's grip on advanced AI hardware.