Meituan identified the high-traffic Owl Alpha model as LongCat-2.0, a 1.6 trillion-parameter MoE model released under the MIT license.
The notable claim is infrastructure: Meituan says training and inference ran on a 50,000-card Chinese-chip cluster, with a one-million-token context window and only part of the model active per token.
What to watch
The story matters because it turns a technical product move into a business and policy signal. The next test is adoption outside the launch circle and whether the numbers hold up in daily use.
Sources: Meituan, GitHub, Hugging Face, OpenRouter, CocoLoop.