Instructions to use local-inference-lab/GLM-5.3-Flash-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use local-inference-lab/GLM-5.3-Flash-NVFP4 with Transformers:

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

pipe = pipeline("text-generation", model="local-inference-lab/GLM-5.3-Flash-NVFP4")
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)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("local-inference-lab/GLM-5.3-Flash-NVFP4")
model = AutoModelForMultimodalLM.from_pretrained("local-inference-lab/GLM-5.3-Flash-NVFP4", 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=256)
print(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use local-inference-lab/GLM-5.3-Flash-NVFP4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "local-inference-lab/GLM-5.3-Flash-NVFP4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "local-inference-lab/GLM-5.3-Flash-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/local-inference-lab/GLM-5.3-Flash-NVFP4

SGLang

How to use local-inference-lab/GLM-5.3-Flash-NVFP4 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 "local-inference-lab/GLM-5.3-Flash-NVFP4" \
    --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": "local-inference-lab/GLM-5.3-Flash-NVFP4",
		"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 "local-inference-lab/GLM-5.3-Flash-NVFP4" \
        --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": "local-inference-lab/GLM-5.3-Flash-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use local-inference-lab/GLM-5.3-Flash-NVFP4 with Docker Model Runner:
```
docker model run hf.co/local-inference-lab/GLM-5.3-Flash-NVFP4
```

Community evaluation (2026-10-03): GSM8K 97.35%, GPQA 88.69% (NV-protocol 5-run mean), MMLU-Pro 83.70% — 4× RTX PRO 6000 Blackwell, Karmic Kraken serving stack

by malaiwah - opened 7 days ago

Discussion

malaiwah

7 days ago

Adds a community-evaluation section to the model card with independent accuracy measurements of this checkpoint (commit 175ae8ce), served end-to-end on 4× RTX PRO 6000 Blackwell (96 GB each) with the Karmic Kraken vLLM stack (ghcr.io/local-inference-lab/vllm:karmic-kraken-beta, digest-pinned, vLLM 0.1.dev22021+g93dabce32, MTP draft 3, LMCache engine-driven L1 64 GiB / L2 512 GiB).

Highlights (llm-inference-bench v0.7.6, pinned sha256-verified datasets, per-item scoring, Wilson 95% CIs):

GSM8K: 97.35% (1284/1319) greedy — at/above the BF16-band community anchors (96.9–97.2%); concurrency-invariant (C24 = C30)
GPQA Diamond: 88.69% 5-run mean at the NVIDIA sibling-table protocol (temp 1.0 / top_p 0.95 / 327,680 tokens); temp-0 greedy-loop truncations shown to be a sampling artifact
MMLU-Pro: 83.70% (837/1000) greedy — first published MMLU-Pro datapoint for this checkpoint

Full methodology, per-run numbers, and confidence intervals in the README diff. Raw JSON outputs available on request.

malaiwah

7 days ago

Closing: this discussion was created without a git reference (empty PR). The complete evaluation has been submitted as PR #6 with the README diff: https://huggingface.co/local-inference-lab/GLM-5.3-Flash-NVFP4/discussions/6

malaiwah changed discussion status to closed 7 days ago

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