Meituan Open-Sources LongCat-2.0: 1.8T Params Trained on Domestic Chips

multimodalart · x · 2026-07-05

Meituan has released its LongCat-2.0 large language model with open weights, boasting a parameter scale of around 1.8T (1T+).

A major selling point is that it was trained entirely on domestically produced ASICs, marking a milestone case of local computing power supporting the training of trillion-parameter models.

Related event: Meituan Open-Sources LongCat-2.0 on Domestic Chips(3 posts)→

Original post →

More from Infra

Infra channel →