Instructions to use MiniMaxAI/MiniMax-M3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MiniMaxAI/MiniMax-M3 with Transformers:

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

pipe = pipeline("image-text-to-text", model="MiniMaxAI/MiniMax-M3", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("MiniMaxAI/MiniMax-M3", trust_remote_code=True)
model = AutoModelForMultimodalLM.from_pretrained("MiniMaxAI/MiniMax-M3", trust_remote_code=True)
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use MiniMaxAI/MiniMax-M3 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MiniMaxAI/MiniMax-M3"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MiniMaxAI/MiniMax-M3",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/MiniMaxAI/MiniMax-M3

SGLang

How to use MiniMaxAI/MiniMax-M3 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 "MiniMaxAI/MiniMax-M3" \
    --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": "MiniMaxAI/MiniMax-M3",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "MiniMaxAI/MiniMax-M3" \
        --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": "MiniMaxAI/MiniMax-M3",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use MiniMaxAI/MiniMax-M3 with Docker Model Runner:
```
docker model run hf.co/MiniMaxAI/MiniMax-M3
```

Can MiniMax-M3 run on 2x NVIDIA DGX Spark 4TB?

#16

by abosk - opened 2 days ago

Discussion

abosk

2 days ago

Hi,

I’m evaluating whether MiniMax-M3 can run locally on 2x NVIDIA DGX Spark units, each with 128 GB unified memory and 4 TB NVMe storage.

From what I understand, vLLM seems to be the only inference backend that could realistically work well on DGX Spark. Could you confirm whether MiniMax-M3 can run on this setup using vLLM, and whether this hardware would be enough for practical inference?

Thanks in advance.

south-american-cowboy

about 21 hours ago

4-bit quantized it should be able to. I am running thru this experiment right now... using llama.cpp with unsloth minimax-m3 model.

llmfan46

about 15 hours ago

•

edited about 5 hours ago

4-bit quantized it should be able to. I am running thru this experiment right now... using llama.cpp with unsloth minimax-m3 model.

Not possible.

2x NVIDIA DGX Spark with 128GB of Unified Memory = 256 GB of Unified Memory

MiniMax-M3 Q4_K_M = 257 GB

MiniMax-M3 Q4_K_S = 241 GB

There is not enough headroom.

One of the Q3 quants might be possible though, like Q3_K_M = 203 GB or Q3_K_S = 184 GB, Q3_K_L = 220 GB might be possible if you use low settings for context length.

alkeryn

about 7 hours ago

we realy need better more affordable hardware seriously.

llmfan46

about 7 hours ago

we realy need better more affordable hardware seriously.

Yes but how?

Right now people are buying the RTX 5090 for up to 5K for 32GB of VRAM, the RTX PRO 6000 now has an MSRP of 13500 (which does not have any NVLINK support btw) and even with 2x RTX PRO 6000 (total cost 27000 at MSRP) and even at that price it still isn't even enough for this model at Q4.

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