Meituan Releases LongCat-2.0, a 1.6-Trillion-Parameter Agentic Coding Model Trained Entirely on Chinese Chips
The AMW Read
Novelty 2: Meituan is a known player but the model's training on domestic chips and 1.6T scale updates the CN open-weight strategy; significance 2: impacts AI coding and sovereign compute segments.
Meituan Releases LongCat-2.0, a 1.6-Trillion-Parameter Agentic Coding Model Trained Entirely on Chinese Chips
Chinese food-delivery and services giant Meituan has open-sourced LongCat-2.0, a 1.6-trillion-parameter agentic coding model released under the MIT license. The company states that the model, which has been leading on the OpenRouter inference platform, was trained entirely on domestically produced Chinese semiconductors, marking a significant milestone in China's effort to build frontier AI capability without reliance on advanced Western chips.
Why it matters: LongCat-2.0 updates the structural forces around compute sovereignty and the capital-compression arc for open-weight foundation models. The model's training on Chinese-made chips demonstrates an end-to-end domestic AI stack — from hardware through training to deployment — directly responding to US export controls that have restricted Chinese access to NVIDIA's most advanced GPUs. By open-sourcing a 1.6T-parameter model under a permissive MIT license, Meituan also advances the pattern of hyperscaler-distribution moat-building: the company can amortize massive training costs across internal infrastructure, while using the open release to attract developer mindshare and contributions, similar to Meta's strategy with Llama but with an added geopolitical signal.
For the broader AI market, LongCat-2.0 validates that Chinese labs can produce competitive large models on domestic silicon, potentially accelerating sovereign AI initiatives in other regions and intensifying the debate about whether export controls are effective in slowing China's frontier-AI progress. The model's reported OpenRouter leadership further suggests that agentic coding — AI-assisted software development — is becoming a proving ground for model capability, a segment where Chinese open-weight entrants are increasingly competitive with Western proprietary offerings.
