Instructions to use beyoru/seul-preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use beyoru/seul-preview with Transformers:

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

pipe = pipeline("text-generation", model="beyoru/seul-preview")
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("beyoru/seul-preview")
model = AutoModelForMultimodalLM.from_pretrained("beyoru/seul-preview")
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 beyoru/seul-preview with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/beyoru/seul-preview

SGLang

How to use beyoru/seul-preview 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 "beyoru/seul-preview" \
    --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": "beyoru/seul-preview",
		"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 "beyoru/seul-preview" \
        --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": "beyoru/seul-preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use beyoru/seul-preview with Docker Model Runner:
```
docker model run hf.co/beyoru/seul-preview
```

seul

Intelligence grows by accumulation, not replacement.

seul is the model in a new family of agentic language models, designed for long-horizon reasoning and reliable business tool use.

Rather than optimizing only for benchmark performance, seul is trained to maintain context across extended workflows, interact safely with enterprise tools, and improve through reinforcement learning with verifiable outcomes.

seul

Design Goals

Long-horizon reasoning
Reliable enterprise tool use
Stable multi-turn planning
Verifiable execution
Efficient reinforcement learning

Results — AutomationBench (60-task subset, self-hosted)

Model	Size	Partial credit	Strict pass
Ornith-1.0-9B	9B	22%	3%
seul-preview	9B	28%	7%
Qwen3.6-27B	27B	34%	10%

RLVR training raises partial credit +6 points and more than doubles the strict pass rate over the base, closing a large fraction of the gap to a 3× larger model on the same tasks.

Limittion

This training process was conducted on only one domain.
Due to limited GPU availability, the model was trained for only about half of the originally planned training steps, which may have prevented it from reaching its full potential.

Citation

If you use seul in your research or projects, please cite:

@misc{seul2026,
  title        = {seul-preview},
  author       = {beyoru},
  year         = {2026},
  publisher    = {Hugging Face},
  howpublished = {\url{https://huggingface.co/beyoru/seul-preview}}
}

Downloads last month: 580

Safetensors

Model size

9B params

Tensor type

BF16

Collection including beyoru/seul-preview

Agent

Collection

5 items • Updated about 6 hours ago