MEITUAN-W Open-Sources LongCat-2.0 Trillion-Parameter Large Model
I'm LongbridgeAI, I can summarize articles.MEITUAN-W+1.650 (+2.304%) Short selling $456.75M; Ratio 18.922% announced the official open-sourcing of its trillion-parameter large model LongCat-2.0, while simultaneously releasing inference code for domestic AI chips. LongCat-2.0 has total parameters of 1.6 trillion, with average activation of about 48 billion. The architecture introduces LongCat sparse attention and N-gram Embedding, improving long-context processing efficiency and token-level representation capability, while combining dynamic activation to further strengthen code understanding, generation and execution performance.(HK stocks quote is delayed for at least 15 mins.Short Selling Data as at 2026-07-03 16:25.)
MEITUAN-W (03690.HK) +1.650 (+2.304%) Short selling $456.75M; Ratio 18.922% announced the official open-sourcing of its trillion-parameter large model LongCat-2.0, while simultaneously releasing inference code for domestic AI chips.
LongCat-2.0 has total parameters of 1.6 trillion, with average activation of about 48 billion. The architecture introduces LongCat sparse attention and N-gram Embedding, improving long-context processing efficiency and token-level representation capability, while combining dynamic activation to further strengthen code understanding, generation and execution performance.
(HK stocks quote is delayed for at least 15 mins.Short Selling Data as at 2026-07-03 16:25.)