peonist-ai / halogen-server

The fastest way to run Qwen3.8 27B on Strix Halo (gfx1151)

buqi-code / buqi-minimax-h3-multigpu

MiniMax-H3 multi-GPU parallel inference for ComfyUI | 多卡并行加速节点:2-8 GPU Ulysses sequence parallel, bit-identical video+audio generation