LongCat-2.0 is Meituan's newly released and open-sourced next-generation trillion-parameter model, with 1.6T total parameters and about 48 billion parameters activated per token. The model was trained entirely on domestic chips and natively supports million-token context lengths. LongCat-2.0 has been systematically optimized for long-text understanding, code comprehension, repository-level edits, automated task execution, and multi-turn tool use. It processes long-context information more efficiently and delivers stable, strong performance in development collaboration, agent workflows, and complex reasoning tasks.
Used with Tabbit Browser, LongCat-2.0's 1M ultra-long context means conversations no longer lose their memory. Whether you are following up on technical details, digging into a long article, or discussing a complex issue back and forth, the context stays coherent so your thinking is not interrupted.
Tabbit has integrated LongCat-2.0 right away. Click "Switch Model" in the lower-right corner of the chat box to try it.
If you have not tried Tabbit yet, visit www.tabbit.com or click "Read Original" to enter the official website and download it for free.