Instructions to use AlexanderKyng/Qwen3.6-27B-MTP-48Gb with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use AlexanderKyng/Qwen3.6-27B-MTP-48Gb with Transformers:

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

pipe = pipeline("image-text-to-text", model="AlexanderKyng/Qwen3.6-27B-MTP-48Gb")

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("AlexanderKyng/Qwen3.6-27B-MTP-48Gb", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use AlexanderKyng/Qwen3.6-27B-MTP-48Gb with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "AlexanderKyng/Qwen3.6-27B-MTP-48Gb"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "AlexanderKyng/Qwen3.6-27B-MTP-48Gb",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/AlexanderKyng/Qwen3.6-27B-MTP-48Gb

SGLang

How to use AlexanderKyng/Qwen3.6-27B-MTP-48Gb 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 "AlexanderKyng/Qwen3.6-27B-MTP-48Gb" \
    --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": "AlexanderKyng/Qwen3.6-27B-MTP-48Gb",
		"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 "AlexanderKyng/Qwen3.6-27B-MTP-48Gb" \
        --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": "AlexanderKyng/Qwen3.6-27B-MTP-48Gb",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Unsloth Desktop
Docker Model Runner
How to use AlexanderKyng/Qwen3.6-27B-MTP-48Gb with Docker Model Runner:
```
docker model run hf.co/AlexanderKyng/Qwen3.6-27B-MTP-48Gb
```

Downloading!!

by pawarshardul - opened Jul 20

Discussion

pawarshardul

Jul 20

i will report back when i run this !!! thank you for amazing work!!

pawarshardul

Jul 21

this one didnt cause any re starts yet -- but weirdly -- it deleted files on my desktop - i was trying the prompt from https://huggingface.co/sokann/GLM-5.2-GGUF-2.244bpw/discussions/2 for creating llamawatch -- in llama ui -- i was away from desktop -- as soon as i see it -- i stopped its execution but till then the files were gone permenantly -- same issue is reported by user in https://huggingface.co/migtissera/Tess-4-27B/discussions/5 -- so it may not be your models fault -- this may be qwen 3.6's error ---
for now i will stop using qwen models !! need to try to recover the files first!! Thank you !!

AlexanderKyng

Owner Jul 21

Oh wow, I am sorry to read that.
Yes, Qwen3.6-27B (I personally never had such behavior with 35B), is known to easily delete files without warning. That's why many (and myself) advise to use an agent harness that blocks all deletions and ask for the user's agreement. Avoid 'YOLO' modes as well by the way.

I have never seen this behavior with Qwopus3.6-Coder models, they might be less tempted to delete files and folders than base Qwen3.6, but I cannot be certain.

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