Instructions to use shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5 with Transformers:

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

pipe = pipeline("text-generation", model="shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5")
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, AutoModelForImageTextToText

processor = AutoProcessor.from_pretrained("shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5")
model = AutoModelForImageTextToText.from_pretrained("shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5")
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 shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5

SGLang

How to use shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5 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 "shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5" \
    --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": "shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5",
		"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 "shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5" \
        --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": "shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5 with Docker Model Runner:
```
docker model run hf.co/shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5
```

Qwen3.6-35B-A3B PARO full4096-e5 — legacy/original format

This is the original ParoQuant export for Qwen/Qwen3.6-35B-A3B, using the full4096-e5 calibration run.

Format: legacy/original ParoQuant safetensors export
Quantization: W4A16 ParoQuant, bits=4, group_size=128, krot=8
model.safetensors: 23,284,714,104 bytes
Artifact BPW: 5.3222 using a 35B denominator
Contains the original duplicate fp16 .weight fallback tensors for modules that also have .qweight

A fully packed version with those duplicate fallback tensors removed is available separately at:

shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5-packed

Quality reference

Canonical tx4/quality3 evaluation against the original BF16 HF model:

Model	PPL ↓	ΔNLL ↓	KL nats ↓	Top-1 % ↑
PARO full4096-e5	6.6216	+0.009506	0.034684	92.000

Notes

This artifact requires a ParoQuant-compatible loader/runtime; it is not a plain unquantized Transformers checkpoint.

Downloads last month: 22

Safetensors

Model size

7B params

Tensor type

I32

F16

I16

Model tree for shisa-ai/Qwen3.6-35B-A3B-PARO-full4096-e5

Base model

Qwen/Qwen3.6-35B-A3B

Quantized

(420)

this model