DeepSeek V4 Flash 284B (13B active)
on NVIDIA DGX Spark · llama.cpp · 131,072 ctx
Oct 3, 2026
Summary
User reports DeepSeek-V4-Flash-0731 at 19.3 to 19.7 tokens/second single-stream on one NVIDIA DGX Spark (GB10, 128GB unified memory).
Setup is llama.cpp build 10235 with Unsloth UD-IQ2_M dynamic 2-bit weights, 131072 tokens of context per slot, flash attention on, and no speculative decoding.
Prefill runs 200 to 370 tok/s depending on request shape; four concurrent slots give 52 tok/s aggregate.