Summary
Meituan open-sourced LongCat-2.0 on June 30, 2026 — a 1.6-trillion-parameter MoE model with a 1M-token context window that the company claims is the first model of its scale trained end-to-end on a 50,000-chip domestic Chinese ASIC cluster. The release positions Meituan as proof that frontier-scale AI can be built without Nvidia hardware, directly challenging U.S. export controls. The model was previously known as the anonymous “Owl Alpha” leader on OpenRouter.
Source Analysis
Primary source: The Next Web, June 30, 2026. Corroboration available from VentureBeat and explainx.ai.
PreScreening Notes
Score: 2 — Duplicate of an item already in the pipeline (pipeline/3-evaluated/2026-06-30-meituan-longcat-2-domestic-chips.md). Same source URL and story; stale re-ingest left in 1-incoming/ after prior prescreening. Rejected to prevent parallel tracking.