Instructions to use jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4
# Run inference directly in the terminal:
llama cli -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4
# Run inference directly in the terminal:
llama cli -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4
# Run inference directly in the terminal:
./llama-cli -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4
# Run inference directly in the terminal:
./build/bin/llama-cli -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

Use Docker

docker model run hf.co/jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

LM Studio
Jan
Ollama
How to use jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF with Ollama:
```
ollama run hf.co/jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4
```
Unsloth Desktop

How to use jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF with Docker Model Runner:
```
docker model run hf.co/jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4
```

Lemonade

How to use jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

Run and chat with the model

lemonade run user.NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF-NVFP4

List all available models

lemonade list

Hermes Agent

How to use jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF:NVFP4" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

NVIDIA Nemotron-3-Super-120B-A12B-NVFP4 (GGUF)

GGUF conversion of nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4 via unsloth/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4.

Model Details

Architecture: Hybrid Mamba2-Transformer LatentMoE (NemotronH)
Parameters: 120B total, 12B active per token
Experts: 512 routed (top-22) + 1 shared
Layers: 88 (interleaved Mamba2, MoE, and Attention)
Quantization: NVFP4 (FP4 E2M1 with UE4M3 block scales)
GGUF Size: ~64GB

Quantization Format

Weight tensors use GGML_TYPE_NVFP4:

FP4 E2M1: 4-bit float (values: 0, ±0.5, ±1, ±1.5, ±2, ±3, ±4, ±6)
Block structure: 64 elements per block = 4 UE4M3 scale bytes + 32 packed nibble bytes (36 bytes total)
1D tensors (norms, biases): stored as F32

How to Use

Requires Salamander (llama.cpp fork with NVFP4 and NemotronH LatentMoE support).

./salamander-cli -m NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-NVFP4.gguf -p "Hello" -ngl 99

Conversion

Converted using convert_hf_to_gguf.py with native NVFP4 output:

python convert_hf_to_gguf.py /path/to/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4 --outtype nvfp4

The converter dequantizes the original ModelOpt FP4/FP8 weights to float, then re-quantizes to GGML's native NVFP4 block format.

License

NVIDIA Open Model License

Downloads last month: 234

GGUF

Model size

121B params

Architecture

nemotron_h_moe

Hardware compatibility

4-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for jdziat/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4-GGUF

Base model

nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4

Quantized

(13)

this model