DeepSeek V4.1 Flash 552B (16B active)
on NVIDIA DGX Spark · 32,768 ctx
Summary
User reports DeepSeek-V4.1-Flash at 2.68 tok/s median decode (2.64-2.71 tok/s range) on a single NVIDIA DGX Spark (GB10) with 128 GB unified memory.
Setup uses native FP4 experts, a 73.8 GB arena holding 25.6% of routed experts with the rest streamed from NVMe, DSpark speculative decoding, and MAX_SEQ 32768. Prefill is 337 tok/s on a 5,014-token prompt (14.9 s).
Without DSpark decode is 1.75 tok/s. Expert hit rate is about 0.83 and about 0.9 GB is read from NVMe per token.