Instructions to use nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Local Apps Settings

vLLM

How to use nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16

SGLang

How to use nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 with Docker Model Runner:
```
docker model run hf.co/nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
```

MLX 4-bit quant available (M4 Pro benchmarks included)

#63

by BrendanL79 - opened 25 days ago

Discussion

BrendanL79

25 days ago

MLX 4-bit quant for Apple Silicon:
https://huggingface.co/BrendanL79/Nemotron-3-Nano-30B-A3B-MLX-4bit

Recipe: mlx_lm.convert -q --q-bits 4 --q-group-size 64 (mlx-lm supports nemotron_h out of the box).

Measured on a Mac Mini M4 Pro 24GB, same harness for both models:

metric	Nemotron 3 Nano 30B-A3B 4bit	Qwen3.6-35B-A3B 4bit
decode (1024 tok, greedy)	~95 tok/s	~92 tok/s
peak memory	17.1 GB, flat across generation	18.7 GB, growing
prefill @1K prompt	~620 tok/s	~686 tok/s
prose perplexity (1024-tok window)	2.98	1.45
chat trigram diversity	0.962	0.913

The flat memory curve is the hybrid-Mamba effect — only 6 of 52 layers carry KV cache. Thinking mode is on by default via the chat template; give it ≥512 tokens or it spends the budget reasoning.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment