Meituan releases open-weight LongCat-2.5 agent model
Meituan released the weights of LongCat-2.5, a 1.6-trillion-parameter mixture-of-experts model designed for autonomous, long-running coding tasks. The model supports text, images and video, processes up to one million tokens through its API, and is offered under the MIT license. Meituan says it was trained entirely on Chinese hardware, while its benchmark results and 4-bit quantization accuracy claims have not been independently verified.
LongCat-2.5 combines open weights, multimodal inputs, long-context API access and deployment guidance for Huawei Ascend accelerators, making it a notable release for developers seeking locally hosted coding agents.
Meituan releases a 1.6-trillion-parameter agent model that never touched an Nvidia GPU
Jonathan Kemper View the LinkedIn Profile of Jonathan Kemper
Nov 15, 2026

ChatGPT-5.6 prompted by THE DECODER
Meituan releases the weights for LongCat-2.5, a 1.6-trillion-parameter model built for AI agents. It was trained entirely on Chinese hardware and performs at the level US models like Claude Opus 5 reached a few months ago on coding benchmarks.
#LongCat-2.5#Meituan#Huawei Ascend#Huawei Atlas A2#Claude Opus 5#GPT-5.6 Sol#Deepseek-V4.1-Flash