Chinese technology giant Meituan has introduced LongCat-2.0, a 1.6-trillion-parameter open-source large language model that the company claims is the first frontier-scale AI model trained and deployed entirely on domestically developed Chinese AI chips. The model was built using a 50,000-chip domestic compute cluster, marking a significant milestone in China’s drive toward AI hardware self-reliance.

LongCat-2.0 supports a 1 million-token context window and is designed for demanding enterprise workloads such as agentic coding, long-document understanding, and autonomous AI workflows. According to Meituan, the model delivers performance comparable to leading proprietary LLMs while eliminating dependence on foreign AI accelerators.

The announcement comes as China continues investing heavily in domestic AI infrastructure following U.S. export restrictions on advanced semiconductor technologies. If these claims are validated, LongCat-2.0 could become an important proof point that frontier-scale AI training is increasingly possible without NVIDIA GPUs, potentially reshaping the global AI infrastructure landscape.

Happy Learning!!

References

• The Next Web – China’s Meituan says its new AI model was trained on domestic chips
• Reuters – China’s Meituan says new AI model trained on domestic chips