NVIDIA Nemotron-3-Super-120B-A12B
serving recipe.

Floors are the minimum the vLLM recipe publishes across the whole replica, from vLLM's own runs. The control plane compares them with its own sizing from the model config.

• Tensor ParallelFits inside one node's interconnect. single_node_tp
• Multi-Node Tensor ParallelSpans nodes. multi_node_tp
• Multi-Node TP + Pipeline ParallelSpans nodes. multi_node_tp_pp
• Multi-Node Tensor + Expert ParallelSpans nodes. multi_node_tep
• Multi-Node Data + Expert ParallelSpans nodes. multi_node_dep

• B200 · 8 GPUs
• DGX Spark (GB10) · 1 GPU
• DGX Station (GB300) · 1 GPU
• H100 · 8 GPUs
• H200 · 8 GPUs
• RTX Pro 6000 · 1 GPU

Wingsmith runs models through NVIDIA Dynamo, so it serves on NVIDIA GPUs. Other vendors are listed as the recipe states them.

vllm serve nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16 \
  --trust-remote-code \
  --kv-cache-dtype fp8 \
  --enable-auto-tool-choice \
  --tool-call-parser qwen3_xml \
  --reasoning-parser nemotron_v3

Features

• Tool callingOn by defaultQwen3 XML tool-call parser with automatic tool choice
• ReasoningOn by defaultBuilt-in nemotron_v3 reasoning parser (vLLM >= 0.17.1)
• Speculative decodingOpt-inMTP speculative decoding for accelerated inference

1. Register the weights

   In the console, or through the API. Resolving a slug proposes an artifact and writes nothing.

   curl "http://localhost:8000/v1/artifacts/resolve?source=hf:nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16"

2. Read the draft

   The draft carries this recipe as upstream evidence: the tested flags, the precision and memory floor, the parallelism layouts, and the hardware vLLM verified, with links back to the recipe. Nothing from it is applied without you, and the draft warns when the plane's own sizing and the published floor differ by a quarter or more.

3. Create the workload

   Choose the engine, device shape and scale. The plane picks the clusters, and each cluster's operator renders the deployment as a Dynamo graph.

4. Give it an endpoint

   One model name for clients, routed by the gateway to deployments on every cluster serving it.

Run the control plane locally →

• How much GPU memory does NVIDIA Nemotron-3-Super-120B-A12B need?The vLLM recipe publishes a minimum of 75 GB across the whole replica for its NVFP4 variant; BF16 298 GB, FP8 149 GB, BF16 298 GB. That is vLLM's own measurement, not a per-GPU figure.
• Which parallelism layouts does the NVIDIA Nemotron-3-Super-120B-A12B recipe support?Tensor Parallel, Multi-Node Tensor Parallel, Multi-Node TP + Pipeline Parallel, Multi-Node Tensor + Expert Parallel and Multi-Node Data + Expert Parallel.
• Can Wingsmith serve NVIDIA Nemotron-3-Super-120B-A12B?Wingsmith serves models through NVIDIA Dynamo on clusters with NVIDIA GPUs. Register nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16 and the plane attaches this vLLM recipe to the draft as evidence: its tested flags, precision and memory floor, for you to review rather than applied on your behalf.
• Does NVIDIA publish a Dynamo recipe for NVIDIA Nemotron-3-Super-120B-A12B?Yes: 10 deployment entries for vLLM, SGLang and TensorRT-LLM, aggregated and disaggregated.

From the vLLM recipe for nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-BF16 ↗ (snapshot at 4bcc189) and NVIDIA Dynamo's recipes at 165d854, Apache 2.0.