Skip to content
AI HOT (Curated Pool)

Moore Threads open-sources MusaCoder, a code LLM fully trained on domestic GPUs

摩尔线程开源 MusaCoder 代码大模型,9B/27B 参数基于国产 GPU 全链路训练

Moore Threads released MusaCoder, a code model for GPU kernel generation, in 9B and 27B sizes. The full post-training pipeline ran on a domestic MTT S5000 cluster. It auto-generates high-performance CUDA/MUSA kernels from PyTorch ops. On KernelBench, the 27B RL version hits 93.2% Overall Pass@8, beating Claude Opus 4.7 and DeepSeek-V4 Pro per the official report. Models and paper are public.

Why it matters: Moore Threads open-sourced MusaCoder in 9B and 27B versions that convert standard PyTorch operators into CUDA or MUSA native code, with the full training run on its own MTT S5000 cluster.

Read the original ↗Export Markdown