Meituan announced on July 6 that its next-generation trillion-parameter large model, LongCat-2.0, was open-sourced.
According to Jin10, Moore Threads said it has completed adaptation work for the model based on its MTT S5000 AI training-and-inference integrated full-function GPU computing card and the MUSA software stack.
Moore Threads said the adaptation covered end-to-end steps including model loading, launching the inference engine, optimization of key operators, deployment verification, and accuracy validation. The company said this enables LongCat-2.0 to run stable and efficient inference on the MTT S5000.
AI TRENDS | Moore Threads Completes Adaptation for Meituan’s Open-Sourced LongCat-2.0 Model
2026-07-06 03:40:34
Disclaimer:
1. The information provided does not constitute investment advice. Investors should make independent decisions and bear all risks themselves.
2. The copyright of this content belongs to the original author. The views expressed herein are solely those of the author and do not represent the stance or position of this website.
