Meituan Open-Sources LongCat-2.0 on Domestic Chips

Meituan has open-sourced LongCat-2.0, a trillion-scale MoE model with open weights and inference code optimized for domestic chips. Reports say it was trained and deployed end-to-end on a 50,000-card domestic computing cluster, with roughly 48 billion activated parameters on average and native million-token context support.

2026-07-05 ~ 2026-07-06 · 3 related posts

• Meituan Open-Sources LongCat-2.0: 1.8T Params Trained on Domestic Chips — multimodalart · 2026-07-05
• Meituan Open-Sources LongCat-2.0: 1.6T MoE Trained on Domestic Chips — 赛博禅心 · 2026-07-05
• Meituan Open-Sources Trillion-Parameter Model LongCat-2.0 — 智东西 · 2026-07-06