Instructions to use UnipatAI/UniScientist-30B-A3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use UnipatAI/UniScientist-30B-A3B with Transformers:

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

pipe = pipeline("text-generation", model="UnipatAI/UniScientist-30B-A3B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("UnipatAI/UniScientist-30B-A3B")
model = AutoModelForCausalLM.from_pretrained("UnipatAI/UniScientist-30B-A3B", device_map="auto")
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use UnipatAI/UniScientist-30B-A3B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "UnipatAI/UniScientist-30B-A3B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "UnipatAI/UniScientist-30B-A3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/UnipatAI/UniScientist-30B-A3B

SGLang

How to use UnipatAI/UniScientist-30B-A3B 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 "UnipatAI/UniScientist-30B-A3B" \
    --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": "UnipatAI/UniScientist-30B-A3B",
		"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 "UnipatAI/UniScientist-30B-A3B" \
        --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": "UnipatAI/UniScientist-30B-A3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use UnipatAI/UniScientist-30B-A3B with Docker Model Runner:
```
docker model run hf.co/UnipatAI/UniScientist-30B-A3B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Introduction

We present UniScientist, an agentic large language model featuring 30 billion total parameters, with only 3 billion activated per token. Developed by UniPat AI, the model is specifically designed for universal scientific research tasks spanning 50+ disciplines. UniScientist achieves state-of-the-art performance across a range of research benchmarks, including FrontierScience-Research, FrontierScience-Olympiad, DeepResearch Bench, DeepResearch Bench II, and ResearchRubrics.

More details can be found in our Blog.

Key Features

Evolving Polymathic Synthesis: A human-LLM collaborative data paradigm that generates research-grade scientific problems across 50+ disciplines, each accompanied by co-evolved rubrics refined through completeness, consistency, and distinguishability checks.
Agentic Research Loop: The model conducts scientific research by iteratively acquiring evidence, deriving formally-justified results, and updating hypotheses via abductive inference, using tools including web_search, google_scholar, page_fetching, and code_interpreter.
Report Aggregation: Given multiple candidate research reports, the model learns to synthesize a consolidated report integrating the best elements, enabling research quality to self-evolve over time.

Download

You can download the model then run the inference scripts in https://github.com/UniPat-AI/UniScientist.

@misc{unipat2026uniscientist,
  title   = {UniScientist: Advancing Universal Scientific Research Intelligence},
  author  = {UniPat AI Team},
  year    = {2026},
  howpublished = {\url{https://github.com/UniPat-AI/UniScientist}}
}

Downloads last month: 11

Safetensors

Model size

31B params

Tensor type

BF16

Model tree for UnipatAI/UniScientist-30B-A3B

Quantizations

2 models