Instructions to use Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8 with Transformers:

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

pipe = pipeline("image-text-to-text", model="Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8")
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("Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8")
model = AutoModelForMultimodalLM.from_pretrained("Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8", 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 Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8",
		"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/Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8

SGLang

How to use Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8 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 "Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8" \
    --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": "Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8",
		"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 "Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8" \
        --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": "Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8",
		"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 Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8 with Docker Model Runner:
```
docker model run hf.co/Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8
```

GLM-5.3-Flash-Derisked-FP8

Weight-level de-risked GLM-5.3-Flash · 320B MoE / 18B active · FP8 deployment build

Built by Blackfrost · Las Vegas, NV

Status

This is a private, pre-release checkpoint. Load and coherent-generation validation has passed. Refusal, quality-retention, tool-calling, and throughput evaluation are still in progress; no final benchmark claim is made in this card.

Why this model exists

Security teams cannot fully evaluate defenses against a model that refuses to exercise the behavior under test. This checkpoint is intended for authorized red teaming, AI-safety research, guardrail evaluation, detection engineering, and controlled adversarial testing.

Refusal behavior has been modified at the weight level. The production transformation recipe is proprietary and is not included.

Specifications


Architecture	`Glm5NextForConditionalGeneration` — multimodal hybrid-attention MoE
Parameters	320B total · approximately 18B active per token · no expert pruning
Precision	Dynamic FP8 E4M3 mixed precision with selected modules preserved at higher precision
Packaging	Hugging Face `safetensors` · 120 shards
Indexed tensor bytes	328,326,771,576 bytes (~305.8 GiB)
Context	1,048,576 positions architectural ceiling
Modalities	Text and vision
Languages	English and Chinese

The original tokenizer, processor, generation configuration, chat template, and multimodal configuration are preserved with the checkpoint.

Lineage

zai-org/GLM-5.3-Flash-BF16
  └─ Blackfrost weight-level de-risking
       └─ GLM-5.3-Flash-Derisked-FP8        ← this repository


Upstream	`zai-org/GLM-5.3-Flash-BF16`
Applied	Blackfrost proprietary weight-level behavior modification and FP8 production conversion
Not applied	Expert pruning

Validation status

Check	Status
Artifact structure and 120-shard index	Passed
Multi-GPU load	Passed
Coherent text generation	Passed
OpenAI-compatible streaming	Passed
Reasoning parser	Passed
Tool-call parser	Passed
Refusal evaluation	In progress
Quality-retention evaluation	Pending
Final throughput publication	Pending

Results will be added only after the exact uploaded revision has been independently verified.

Serving

The internal validation lane uses SGLang with tensor and expert parallelism across eight Blackwell GPUs. GLM-5.3-Flash requires a recent runtime with native glm5_next and FP8 support.

python3 -m sglang.launch_server \
  --model-path /path/to/GLM-5.3-Flash-Derisked-FP8 \
  --served-model-name GLM-5.3-Flash-Derisked-FP8 \
  --trust-remote-code \
  --tp-size 8 \
  --ep-size 8 \
  --reasoning-parser glm45 \
  --tool-call-parser glm47 \
  --host 0.0.0.0 \
  --port 8000

Treat this as a baseline, not a universal launch command. Pin the exact runtime image after validation and tune memory, context, attention backends, and MoE kernels for the target hardware.

License

This repository follows the upstream MIT license. Review the included license and upstream materials before deployment.

Intended use

Authorized offensive-security and red-team research
AI-safety and alignment evaluation
Guardrail, classifier, and detection development
Controlled agent and tool-use testing

Disclaimer

This checkpoint is not a safety-stock model. It may produce content that consumer models decline. Outputs are untrusted and require independent controls, access restrictions, logging, and human review.

The model is provided as-is, without warranty. Evaluation results describe only the exact checkpoint and harness stated; they are not safety guarantees. Further fine-tuning, merging, quantization, pruning, or modification creates a different artifact not covered by this card.

Downloads last month: 3

Safetensors

Model size

321B params

Tensor type

F32

BF16

F8_E4M3

Model tree for Blackfrost-AI/GLM-5.3-Flash-Derisked-FP8

Base model

zai-org/GLM-5.3-Flash-BF16

Quantized

(31)

this model