Meituan releases 1.6-trillion-parameter LLM on Chinese chips

Chinese technology company Meituan has released a massive new artificial intelligence model developed entirely on domestic hardware, marking a significant step in the country's push for technological self-reliance. The company announced LongCat2.0, a 1.6-trillion-parameter open-source large language model, on Tuesday, June 30, 2026.

The model was trained from scratch on a cluster of more than 50,000 domestically produced AI chips, specifically Application-Specific Integrated Circuit (ASIC) superpods. Meituan stated this makes LongCat2.0 the first trillion-parameter model to complete both its demanding pre-training phase and its inference operations on Chinese hardware, a technical milestone for the industry. The company utilized Huawei's Collective Communication Library to coordinate the massive chip cluster during training.

LongCat2.0 is a sparse mixture-of-experts model with a context window of one million tokens, allowing it to process extremely long documents or data streams. The company designed the model to function as a core "brain" for AI agents, with a focus on coding tasks, automated workflow execution, and repository-level software edits. Meituan claims the model has...