Instructions to use Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw with Transformers:

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

pipe = pipeline("image-text-to-text", model="Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw")
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("Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw")
model = AutoModelForMultimodalLM.from_pretrained("Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw", device_map="auto")
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 Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw

SGLang

How to use Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw 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 "Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw" \
    --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": "Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

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 "Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw" \
        --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": "Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw with Docker Model Runner:
```
docker model run hf.co/Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw
```

GLM-5.3-Flash-EXL3-TR3-4bpw / README.md

Mia-AiLab

Update README.md

9eaebb7 verified 11 days ago

preview code

Raw

History Blame Contribute Delete

3.02 kB

metadata

base_model: zai-org/GLM-5.3-Flash
library_name: transformers
pipeline_tag: image-text-to-text
license: other
license_name: shapleymcg-license-1.0
license_link: https://huggingface.co/Mia-AiLab/GLM-5.3-Flash-EXL3-TR3-4bpw/blob/main/LICENSE
tags:
  - shapleymcg
  - glm
  - exl3
  - tr3
  - vllm
  - quantized

GLM-5.3-Flash TR3 4bpw — mirror (2x DGX Spark recipe)

This is not an original quantization. It is a byte-identical redistribution of brandonmusic/GLM-5.3-Flash-tr3-4bpw at Hugging Face revision 5ab363a8dcf6405955fd5f99671e01a1c9fb124b so that the 2× DGX Spark recipe MiaAI-Lab/GLM-5.3-Flash-EXL3-2x-DGX-Sparks stays fetchable if the upstream Hub id moves.

Brandon M. Music created the EXL3/TR3 checkpoint (ShapleyMcg). Z.AI created the base model. Mia's AI Lab only re-hosts this snapshot and serves it on GB10.

@misc{music2026shapleymcg,
  author = {Music, Brandon M.},
  title  = {ShapleyMCG: An Auditable Calibration-to-Encoding Pipeline for
            Low-Bit Mixture-of-Experts Models},
  year   = {2026},
  url    = {https://github.com/brandonmmusic-max/shapleymcg},
  note   = {Licensed under the ShapleyMcg License v1.0}
}

License

This checkpoint / ShapleyMcg work: ShapleyMCG License 1.0 (source-available; not OSI “open source”). The LICENSE file is the upstream file, unmodified.
Base model zai-org/GLM-5.3-Flash: MIT, Copyright (c) 2026 Z.AI Co., Ltd. Keep that notice with any copy of the base work.
Spark serve scripts (separate GitHub repo): MIT.

Do not relicense these shards as MIT. Downstream copies of this Work stay under ShapleyMCG License 1.0.

What this is

Uniform-K4 EXL3/TR3 routed-expert checkpoint of GLM-5.3-Flash (~164 GiB, 120 safetensor shards). Provenance metadata from the upstream snapshot is kept.

This Hub repo is not Brandon’s SM120 B12X / NVFP4-KV / EP2/DCP2 daily driver. Do not use the verdictai/glm53-flash-exl3-k4:… image with a “this is the Spark recipe” assumption. His original model card is saved as ORIGINAL_MODEL_CARD.md.

Spark recipe (2× NVIDIA GB10, TP=2, fp8 MLA KV, DFlash2):

GitHub: https://github.com/MiaAI-Lab/GLM-5.3-Flash-EXL3-2x-DGX-Sparks
Image: ghcr.io/miaai-lab/glm-5.3-flash-2x-dgx-sparks:exl3

DFlash2 (incoai/GLM-5.3-Flash-DFlash2) is a separate checkpoint under CC BY-NC-ND 4.0 and is not mirrored here.

Source pin

Field	Value
Upstream	`brandonmusic/GLM-5.3-Flash-tr3-4bpw`
Upstream revision	`5ab363a8dcf6405955fd5f99671e01a1c9fb124b`
Canonical ShapleyMcg repo	https://github.com/brandonmmusic-max/shapleymcg
Canonical Hub (author)	https://huggingface.co/brandonmusic