Meituan has identified itself as the developer behind LongCat-2.0, a 1.6 trillion-parameter mixture-of-experts model that had been running on OpenRouter for about two months under the alias “Owl Alpha.” The model had been surfacing without attribution before the company publicly claimed it.
The development matters because it adds another large-scale model to an already crowded AI market and highlights how providers are using public model hubs to test or distribute systems before formal announcements. The source also notes that LongCat-2.0 undercuts GPT-5.5 and Claude Sonnet 5 on price by a wide margin, which could influence how developers compare model options.
LongCat-2.0’s appearance on OpenRouter suggests that large model launches may increasingly happen in stages, with some systems effectively operating in the open before their makers step forward. That makes transparency around model identity, performance, and pricing more important for users evaluating which tools to build on.
For readers following AI infrastructure and adjacent crypto markets, the story is less about speculation and more about competition: lower-cost, high-capacity models can change how platforms price access and how developers allocate workloads. It is another example of how fast-moving AI releases are reshaping the software and services landscape.