Instructions to use prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF with Transformers:

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

pipe = pipeline("text-generation", model="prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use prithivMLmods/Llama-3.3-70B-Instruct-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 prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4
# Run inference directly in the terminal:
llama cli -hf prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4
# Run inference directly in the terminal:
llama cli -hf prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4
# Run inference directly in the terminal:
./build/bin/llama cli -hf prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4

Use Docker

docker model run hf.co/prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4

Llama new
LM Studio
Jan

vLLM

How to use prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4

SGLang

How to use prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF 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 "prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF" \
    --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": "prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF",
		"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 "prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF" \
        --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": "prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF with Ollama:
```
ollama run hf.co/prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4
```
Unsloth Desktop

How to use prithivMLmods/Llama-3.3-70B-Instruct-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 prithivMLmods/Llama-3.3-70B-Instruct-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": "prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF with Docker Model Runner:
```
docker model run hf.co/prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4
```

Lemonade

How to use prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4

Run and chat with the model

lemonade run user.Llama-3.3-70B-Instruct-NVFP4-GGUF-NVFP4

List all available models

lemonade list

Hermes Agent

How to use prithivMLmods/Llama-3.3-70B-Instruct-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 prithivMLmods/Llama-3.3-70B-Instruct-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 prithivMLmods/Llama-3.3-70B-Instruct-NVFP4-GGUF:NVFP4

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use prithivMLmods/Llama-3.3-70B-Instruct-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 prithivMLmods/Llama-3.3-70B-Instruct-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 "prithivMLmods/Llama-3.3-70B-Instruct-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"

Llama-3.3-70B-Instruct-NVFP4-GGUF

Llama-3.3-70B-Instruct is Meta’s 70B text-only instruction-tuned model, released in December 2024, with a 128K context window and grouped-query attention. Based on Llama 3.1, it was pretrained on approximately 15T public tokens and aligned using supervised fine-tuning and RLHF for multilingual dialogue and tool use across eight supported languages. Key benchmark scores include 86.0 MMLU (CoT), 68.9 MMLU-Pro, 92.1 IFEval, 50.5 GPQA Diamond, 88.4 HumanEval, 87.6 MBPP EvalPlus, 77.0 MATH, and 91.1 MGSM. It surpasses Llama 3.1 405B Instruct on IFEval, MATH, and GPQA Diamond, while remaining competitive on coding and tool-use benchmarks. Supports Transformers chat templates, original Llama checkpoints, and 8-bit or 4-bit bitsandbytes loading. Released under Meta’s gated Llama 3.3 Community License, with attribution and additional licensing requirements for large-scale services.

Model Files

File Name	Quant Type	File Size	File Link	Description
Llama-3.3-70B-Instruct-NVFP4-00001-of-00005.gguf	NVFP4	8.94 GB	Link	NVFP4 quantization (Shard 1/5). Highly optimized for NVIDIA Blackwell hardware accelerators (e.g., RTX Pro 6000, RTX 50-series, B200, B300), offering native hardware-accelerated FP4 Tensor Core speeds.
Llama-3.3-70B-Instruct-NVFP4-00002-of-00005.gguf	NVFP4	8.88 GB	Link	NVFP4 quantization (Shard 2/5).
Llama-3.3-70B-Instruct-NVFP4-00003-of-00005.gguf	NVFP4	8.93 GB	Link	NVFP4 quantization (Shard 3/5).
Llama-3.3-70B-Instruct-NVFP4-00004-of-00005.gguf	NVFP4	8.88 GB	Link	NVFP4 quantization (Shard 4/5).
Llama-3.3-70B-Instruct-NVFP4-00005-of-00005.gguf	NVFP4	7.1 GB	Link	NVFP4 quantization (Shard 5/5).

NVFP4 Test Run

Tested Llama-3.3-70B-Instruct-NVFP4-00001-of-00005.gguf on an NVIDIA RTX PRO 6000 Blackwell Server Edition GPU with 96 GB of VRAM, using llama-cli with full GPU offloading.

Performance Metrics

Metric	Value
Average Prompt Processing Speed	1,024.93 tokens/s
Average Generation Speed	33.13 tokens/s
Total Wall Time	88.4 seconds
Context Length	8,192 tokens
KV Cache (K)	`q8_0`
GPU	NVIDIA RTX PRO 6000 Blackwell Server Edition
GPU VRAM Capacity	96 GB
Peak GPU VRAM Usage	40.61 GiB
Peak GPU Utilization	99%
GPU Layers Offloaded	99 (`-ngl 99`)
Random Seed	42
Speculative Decoding	No
Test Runs	5

Test Shots

Shot	Test
Shot 1	C: Insert Node at End of Singly Linked List
Shot 2	Jogger and Train Passing Time
Shot 3	Two Pipes Filling a Tank
Shot 4	Students in Examination Rooms
Shot 5	Father and Son Ages

Run Inference

./build/bin/llama-cli \
  -m /path/to/models/Llama-3.3-70B-Instruct-NVFP4-00001-of-00005.gguf \
  -ngl 99 \
  -c 8192 \
  --cache-type-k q8_0 \
  --seed 42 \
  -n 4096 \
  -st \
  -p "Your prompt here"

Performance results reflect this specific hardware and inference configuration. Actual throughput may vary depending on the prompt, model, software build, and runtime settings.