Instructions to use JuicesAI/CTFty-9b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use JuicesAI/CTFty-9b with Transformers:

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

pipe = pipeline("text-generation", model="JuicesAI/CTFty-9b")
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("JuicesAI/CTFty-9b")
model = AutoModelForMultimodalLM.from_pretrained("JuicesAI/CTFty-9b", 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 JuicesAI/CTFty-9b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "JuicesAI/CTFty-9b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "JuicesAI/CTFty-9b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/JuicesAI/CTFty-9b

SGLang

How to use JuicesAI/CTFty-9b 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 "JuicesAI/CTFty-9b" \
    --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": "JuicesAI/CTFty-9b",
		"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 "JuicesAI/CTFty-9b" \
        --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": "JuicesAI/CTFty-9b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use JuicesAI/CTFty-9b with Docker Model Runner:
```
docker model run hf.co/JuicesAI/CTFty-9b
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

CTFty-9b (v2)

CTF/security agent, full SFT from XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B (v1 was trained from Qwen3.5-9B).

Training data — 1651 examples

42 agent runs with verified captured flags (Kaspersky CTF, KubSTU, RCPISS/Avito, НТО — 8 tasks: reverse, pwn, forensics, web, crypto). Runs where the flag was NOT actually captured were excluded.
535 pentest walkthroughs (Pentest-R1)
858 MITRE ATT&CK technique cards
216 MBC malware behavior cards

Recipe

Full fine-tune: 2 epochs, lr 5e-6 cosine, warmup 6 steps, global batch 16 (bs1 x ga16), bf16, max_length 8192, adamw_8bit, gradient checkpointing, RTX PRO 6000.

Train metrics: final logged loss 1.237, avg train_loss 1.267, runtime 71 min.