Nemotron-3 Puzzle 75B-A9B · vLLM · NVFP4 · conc 1

status: done

Configuration

Modelnvidia/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-NVFP4
CompanyNVIDIA
FamilyNemotron
Parameters75.3B / 9.3B (MoE, hybrid Mamba-Transformer)
EnginevLLM
Quant / precisionNVFP4
Why this quantBlackwell-native FP4 (W4A4) — GB10 gets native FP4 compute; FP8 targets Hopper, BF16 (~150 GB) won't fit 121 GB. NVIDIA's own quant, ~44.5 GB. Same Nemotron-3-Super family/quant as the 120B sibling (served on cu130-nightly; older images reject its quant_algo whitelist).
Downloadnvidia/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-NVFP4
Context window65536
Input modalitiestext

Measured results

Prefill tok/s35.0
Decode tok/s19.84
Peak memory (GB)108.43 (system MemAvailable delta (10s sampling) — vLLM 0.85 reservation; real NVFP4 weights ~53.5 GB)
Completed2026-07-11 16:29 +08

Full run command

# vllm/vllm-openai:nightly-aarch64 — loads this NVFP4 checkpoint cleanly (no quant_algo whitelist
# rejection, unlike the older images that blocked the Super-120B sibling). ENTRYPOINT ["vllm","serve"].
scripts/bench-vllm-serving.sh nvidia/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-NVFP4 \
  65536 1 1000 900 256 --trust-remote-code --reasoning-parser nemotron_v3

nvidia-nemotron-3-puzzle-75b-a9b NVIDIA Nemotron NVFP4 41-130B Spark recipe conc-1

Notes

Nemotron-3 Puzzle 75B-A9B (NVFP4) — conc 1 (single-stream). Compressed variant of Nemotron-3-Super (75.3B total / 9.3B active, hybrid Mamba-Transformer MoE: 40 Mamba + 40 MoE + 8 attention blocks). Benchmarked as the DGX-Spark-suitable variant (Blackwell-native FP4).

  • Load: 524 s on nightly-aarch64 (incl. torch.compile cold boot). NVFP4 weights ~53.5 GB (6 shards).
  • Workload: ShareGPT V3, concurrency 1. Hit the 15-min cap at 72/1000 prompts, 0 errors (911.8 s).
  • Throughput (single-stream): prefill 35.0 tok/s, decode 19.84 tok/s; median TTFT 12.8 s. Decode tracks the 9.3B active params — in line with the Super-120B sibling’s single-stream range (~22–24 tok/s for 12B active per the vLLM DGX Spark blog; this 75B/9.3B lands a touch lower at 19.8).