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

Libraries

How to use beyoru/Luna-Ethos with Transformers:

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

pipe = pipeline("text-generation", model="beyoru/Luna-Ethos")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("beyoru/Luna-Ethos")
model = AutoModelForCausalLM.from_pretrained("beyoru/Luna-Ethos")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.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(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use beyoru/Luna-Ethos with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "beyoru/Luna-Ethos"
# 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/Luna-Ethos",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/beyoru/Luna-Ethos

SGLang

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

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

Overview

Luna is a conversational AI model designed for immersive roleplay (RP) and natural chatting. It has been carefully fine-tuned to deliver responses that feel more engaging, expressive, and character-driven than those produced by standard instruction-tuned models. Beyond simply following prompts, it places greater emphasis on tone, personality, and narrative flow—making interactions feel more dynamic and human-like. At the same time, its reasoning capabilities have been further enhanced. It is designed not only to generate stylistically rich responses, but also to think more deeply, connect ideas more coherently, and provide clearer, more structured explanations. This combination allows it to balance creativity with logic—offering answers that are both compelling to read and thoughtfully reasoned.

Ussefull for support task

Defined by character. Guided by values. Expressed through behavior.

Limitations

Optimized primarily for English; multilingual performance is not guaranteed
May occasionally overemphasize tone or style over precision
Not suitable for professional advice (medical, legal, psychological)
Training dataset includes Claude Opus 4.6 (maximum-thinking trajectories) and an RP LoRA merged from the Luna model.
Notes: SFT for reasoning dataset, RL for rp style. Config trainig pretty randomly and not optimize in training.