AI · November 15, 2026 · Moderate impact
Meituan releases LongCat-2.5 agent model trained on Chinese hardware
Meituan has released the weights for LongCat-2.5, a 1.6-trillion-parameter model built for AI agents and trained entirely on Chinese hardware. It scores 88.4 points on Terminal Bench, just behind Claude Opus 5. The sample deployment runs on 194 Huawei Ascend chips, and no Nvidia GPUs are mentioned.
A very large open-weight agent model is competitive on an agentic benchmark while being built and served without Nvidia hardware.