Instructions to use MiniMaxAI/MiniMax-M2.5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MiniMaxAI/MiniMax-M2.5 with Transformers:

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

pipe = pipeline("text-generation", model="MiniMaxAI/MiniMax-M2.5", 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("MiniMaxAI/MiniMax-M2.5", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("MiniMaxAI/MiniMax-M2.5", 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
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use MiniMaxAI/MiniMax-M2.5 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MiniMaxAI/MiniMax-M2.5"
# 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-M2.5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/MiniMaxAI/MiniMax-M2.5

SGLang

How to use MiniMaxAI/MiniMax-M2.5 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-M2.5" \
    --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-M2.5",
		"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 "MiniMaxAI/MiniMax-M2.5" \
        --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-M2.5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use MiniMaxAI/MiniMax-M2.5 with Docker Model Runner:
```
docker model run hf.co/MiniMaxAI/MiniMax-M2.5
```

Which reasoning parser to use?

#33

by g-a-b-y - opened Feb 21

Discussion

g-a-b-y

Feb 21

•

edited Feb 21

The docs indicate we should be running the model with --reasoning-parser minimax_m2_append_think. This works, but causes the <think> tags to be added in the prompt and get displayed in some clients.

A recent issue on vLLM's GitHub says to use the deepseek_r1. Issue here https://github.com/vllm-project/vllm/issues/34625

Can someone from the MiniMax team provide some guidance on this? Maybe there's a bug in the minimax_m2_append_think parser.

AyRick

Feb 21

From what I saw, the minimax_m2_append_think parser works with models that only send the end token and assume that everything is before this token is reasoning.

Minimax M2.5 send both start token and end token, and nothing is sent through reasoning when using the minimax_m2_append_think parser, that is weird, but using the deepseek_r1 works well as it handles models that send both start and end tokens.

Maybe update the documentation as it is useless to have multiple parsers on vLLM that do the same thing

darkstar3537

Apr 4

From what I saw, the minimax_m2_append_think parser works with models that only send the end token and assume that everything is before this token is reasoning.

Minimax M2.5 send both start token and end token, and nothing is sent through reasoning when using the minimax_m2_append_think parser, that is weird, but using the deepseek_r1 works well as it handles models that send both start and end tokens.

Maybe update the documentation as it is useless to have multiple parsers on vLLM that do the same thing

Does deepseek_r1 properly implement interleaved thinking?

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