Homogeneous 2×–4× Spark load; scanner role on 6+ desks. Street llama.cpp Q2 print on one RTX PRO 6000.
Params
unknown
Fastest decode
103 tok/s
2× DGX Spark · custom 2.70 bpw · vLLM
Refusal
Official
Setups
3
Best reported pair · not live serving speed ↗
Runs on
Quoted one-stream decode on real boxes.