Instructions to use yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4 with Transformers:

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

pipe = pipeline("text-generation", model="yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4")

# Load model directly
from transformers import AutoProcessor, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4")
model = AutoModelForImageTextToText.from_pretrained("yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4

SGLang

How to use yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4 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 "yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4 with Docker Model Runner:
```
docker model run hf.co/yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4
```

Mistral-Small-3.2-24B-Instruct-2506 (NVFP4)

This repository contains an NVFP4 quantization of the following base model:

Base model: mistral/Mistral-Small-3.2-24B-Instruct-2506
Quantized model: yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4
Quantization: NVFP4
Quantized with: llmcompressor

What is this?

This is a quantized version of the base model intended to reduce memory usage and improve inference efficiency, while keeping behavior close to the original.

Usage

Add your exact loading snippet here (it depends on how llmcompressor exported the artifacts and which runtime you’re using).

Quantization details

Format: NVFP4
Tooling: llmcompressor
Notes: (add any relevant settings, e.g. target hardware, calibration details, etc.)

Limitations / caveats

Quantized models can differ from the base model in edge cases. If you observe regressions, please compare against the base model and share a minimal repro.

Downloads last month: 20

Safetensors

Model size

14B params

Tensor type

BF16

F32

F8_E4M3

Model tree for yepthatsjason/Mistral-Small-3.2-24B-Instruct-2506-nvfp4

Base model

mistralai/Mistral-Small-3.1-24B-Base-2503

Finetuned

mistralai/Mistral-Small-3.2-24B-Instruct-2506

Quantized

(60)

this model