NVIDIA-Nemotron-3-Super-120B-A12B-FP8

pip install vllm==0.17.1

export MODEL_CKPT=nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-FP8

# Optional: --enable-expert-parallel
vllm serve $MODEL_CKPT \
  --served-model-name nvidia/nemotron-3-super \
  --async-scheduling \
  --dtype auto \
  --kv-cache-dtype fp8 \
  --tensor-parallel-size 4 \
  --max-model-len 262144 \
  --enable-expert-parallel \
  --swap-space 0 \
  --trust-remote-code \
  --gpu-memory-utilization 0.9 \
  --max-cudagraph-capture-size 128 \
  --enable-chunked-prefill \
  --mamba-ssm-cache-dtype float16 \
  --reasoning-parser nemotron_v3 \
  --enable-auto-tool-choice \
  --tool-call-parser qwen3_coder