SN50 Runs the Fastest MiniMax Speeds in the World

TL;DR At RAISE Summit 2026, SambaRack SN50 ran the fastest MiniMax M2.7 inference in the world, as benchmarked by Artificial Analysis. The setup is heterogeneous and disaggregated: one NVIDIA H200 rack (four GPUs) for prefill, one SambaRack SN50 with 16 RDU chips for decode. Decode speeds reach up to 850 tokens per second on short-context workloads and over 450 tokens per second on long-context workloads.