Instructions to use ArliAI/GLM-4.5-Air-Derestricted with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ArliAI/GLM-4.5-Air-Derestricted with Transformers:

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

pipe = pipeline("text-generation", model="ArliAI/GLM-4.5-Air-Derestricted")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ArliAI/GLM-4.5-Air-Derestricted")
model = AutoModelForCausalLM.from_pretrained("ArliAI/GLM-4.5-Air-Derestricted")
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use ArliAI/GLM-4.5-Air-Derestricted with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ArliAI/GLM-4.5-Air-Derestricted"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ArliAI/GLM-4.5-Air-Derestricted",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ArliAI/GLM-4.5-Air-Derestricted

SGLang

How to use ArliAI/GLM-4.5-Air-Derestricted 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 "ArliAI/GLM-4.5-Air-Derestricted" \
    --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": "ArliAI/GLM-4.5-Air-Derestricted",
		"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 "ArliAI/GLM-4.5-Air-Derestricted" \
        --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": "ArliAI/GLM-4.5-Air-Derestricted",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ArliAI/GLM-4.5-Air-Derestricted with Docker Model Runner:
```
docker model run hf.co/ArliAI/GLM-4.5-Air-Derestricted
```

Request for a different model

by ggnoy - opened Nov 24, 2025

Discussion

ggnoy

Nov 24, 2025

Hi, sorry to shit up the discussions page, but could you try this on gemma 3 27b? I'd really like to see if it could be salvaged with this technique. It has really good world knowledge for its size and image analysis capabilities.

jasont2x

Nov 24, 2025

Not the 27b version you are asking for, but there are 12b versions...

https://huggingface.co/collections/grimjim/highlighted-work

s1arsky

Nov 24, 2025

good abliteration of gemma 27B is by mlabonne's. But first version, not second. At least worked for RP.

AutisticPancake

Nov 24, 2025

Out of curiosity, I wonder how would this method work with ServiceNow-AI/Apriel-1.5-15b-Thinker - that one puts GPT-OSS to shame when it comes to refusals, being ridiculously over-censored.

XPforever

Dec 5, 2025

Would love to see a full model of the latest version with abliteration!
https://huggingface.co/zai-org/GLM-4.6

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