← Back to Model Beat
Hardware·Jun 30·all news from June 30, 2026

Meituan's LongCat-2.0 shows China can train massive AI models without Nvidia

Chinese company Meituan has successfully trained a 1.6 trillion parameter artificial intelligence model using only domestically produced hardware. This achievement demonstrates that Chinese firms are increasingly capable of developing large-scale machine learning systems without relying on high-end Nvidia chips, which have been subject to strict U.S. export controls. By bypassing these trade restrictions, the development marks a significant shift in the competitive landscape, showing that indigenous technology can sustain advanced model training despite limited access to global silicon markets.

ModelsLongCat 2.0

Covered by 2 sources

Related stories

ModelsMeituan Releases LongCat-2.0: A 1.6T-Parameter Open MoE Model with Native 1M Context and LongCat Sparse AttentionJul 5