LongCat 2.0
LongCat 2.0 is a 1.6 trillion-parameter open-weight language model developed by meituan. Released in July 2026, it is notable for being trained without nvidia hardware, marking a significant milestone in China’s domestic AI infrastructure capabilities.
Key Specifications & Developments
- Parameters: 1.6 Trillion
- Developer: meituan (Chinese tech conglomerate)
- Hardware Independence: Trained on non-Nvidia infrastructure, demonstrating viability of alternative compute stacks for large-scale model training.
- Status: Open-weight release; achieved top-tier performance benchmarks upon release.
Context & Significance
The release of LongCat 2.0 highlights the decoupling of advanced AI development from US-based semiconductor supply chains. By leveraging domestic or alternative hardware solutions, Meituan has demonstrated that competitive large language models can be produced outside the traditional nvidia-centric ecosystem.