Meituan releases LongCat-2.0 language model

Meituan, the Chinese super-app best known for food delivery and local services, has released LongCat-2.0, a text-focused large language model now published on Hugging Face. The release continues the company's expansion beyond its core commerce business into foundation models.

The LongCat line is part of a growing wave of releases from Chinese technology firms that have made their models openly downloadable rather than gating them entirely behind APIs. As with many of these efforts, LongCat-2.0 ships under a custom license rather than a standard permissive one, so teams evaluating it for production should read the terms closely before building on top of it.

Why it matters

Meituan is not a traditional AI lab, and its move to publish model weights signals how broadly large-platform companies now see in-house language models as strategic infrastructure. For developers, each new openly available model widens the pool of options outside the handful of dominant Western and Chinese labs.

• The model targets text generation, Meituan's stated primary modality.
• It is distributed openly through Hugging Face for download and inspection.
• Licensing is custom, which carries implications for commercial use.

Detailed specifications — including parameter count, context length, and benchmark results — were not part of the initial record, so the model card itself remains the authoritative reference for anyone planning to test it.