Moore Threads Completes Day-0 Adaptation of MiniMax M2.7 Large Model, Supporting MTT S5000 AI Training and Inference GPU

Moore Threads recently completed the Day-0 adaptation of the MiniMax M2.7 large model, which is the industry's first model with deep self-evolution capabilities. It can independently build Agent Harness and complete complex tasks or even participate in its own iteration through collaboration.

This adaptation supports the MTT S5000 flagship AI training and inference GPU, based on the 4th generation MUSA architecture. It has a single-card AI computing power of 1000 TFLOPS, 80GB memory, and fully supports full-precision computing. Moore Threads has also adapted several domestic large models like GLM-5.