Instructions to use Avesed/Qwen3.6-27B-W4A8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Avesed/Qwen3.6-27B-W4A8 with Transformers:

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

pipe = pipeline("text-generation", model="Avesed/Qwen3.6-27B-W4A8")
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("Avesed/Qwen3.6-27B-W4A8")
model = AutoModelForMultimodalLM.from_pretrained("Avesed/Qwen3.6-27B-W4A8")
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Avesed/Qwen3.6-27B-W4A8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Avesed/Qwen3.6-27B-W4A8"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Avesed/Qwen3.6-27B-W4A8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Avesed/Qwen3.6-27B-W4A8

SGLang

How to use Avesed/Qwen3.6-27B-W4A8 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 "Avesed/Qwen3.6-27B-W4A8" \
    --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": "Avesed/Qwen3.6-27B-W4A8",
		"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 "Avesed/Qwen3.6-27B-W4A8" \
        --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": "Avesed/Qwen3.6-27B-W4A8",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Avesed/Qwen3.6-27B-W4A8 with Docker Model Runner:
```
docker model run hf.co/Avesed/Qwen3.6-27B-W4A8
```

Qwen3.6-27B-W4A8

W4A8 quantization of Qwen/Qwen3.6-27B: int4 group-128 weights + int8 dynamic per-token activations (GPTQ via llm-compressor scheme="W4A8").

Why W4A8

int4 weight bandwidth (fast decode) + int8 tensor-core compute (fast prefill) — the best serving quant on the NVIDIA Ampere line (A100 / RTX 3090).

Serving on Ampere (RTX 3090 / A100)

vLLM gates its W4A8 kernels to Hopper. On Ampere the Marlin kernel can run W4A8-int8 but needs a small enablement patch — use vllm-ampere-optimized (prebuilt wheel + Docker image, or the standalone hot-patch). On Hopper it runs out of the box.

Throughput (2× RTX 3090, vLLM tp2, 1024-in / 1024-out)

concurrency	output tok/s	median TTFT	median TPOT
1 (single-user)	46.8	0.84 s	19.8 ms
32 (saturated)	416	14.4 s	63.6 ms

Peak VRAM ~22.8 GiB/card. Single-user ~47 tok/s with sub-second TTFT; saturates ~416 tok/s aggregate.

Downloads last month: 23

Safetensors

Model size

28B params

Tensor type

BF16

Model tree for Avesed/Qwen3.6-27B-W4A8

Base model

Qwen/Qwen3.6-27B

Quantized

(487)

this model