Instructions to use Neura-Tech-AI/Neuron-6x4B-Instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Neura-Tech-AI/Neuron-6x4B-Instruct with Transformers:

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

pipe = pipeline("text-generation", model="Neura-Tech-AI/Neuron-6x4B-Instruct")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Neura-Tech-AI/Neuron-6x4B-Instruct")
model = AutoModelForCausalLM.from_pretrained("Neura-Tech-AI/Neuron-6x4B-Instruct", 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 Neura-Tech-AI/Neuron-6x4B-Instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Neura-Tech-AI/Neuron-6x4B-Instruct"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Neura-Tech-AI/Neuron-6x4B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Neura-Tech-AI/Neuron-6x4B-Instruct

SGLang

How to use Neura-Tech-AI/Neuron-6x4B-Instruct 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 "Neura-Tech-AI/Neuron-6x4B-Instruct" \
    --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": "Neura-Tech-AI/Neuron-6x4B-Instruct",
		"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 "Neura-Tech-AI/Neuron-6x4B-Instruct" \
        --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": "Neura-Tech-AI/Neuron-6x4B-Instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Neura-Tech-AI/Neuron-6x4B-Instruct with Docker Model Runner:
```
docker model run hf.co/Neura-Tech-AI/Neuron-6x4B-Instruct
```

Neuron-6x4B-Instruct

File size: 2,483 Bytes

94a4e8e

Neuron-6x4B-Instruct

Copyright (c) 2026 Neura Tech AI

Neuron-6x4B-Instruct is an open-source Mixture of Experts (MoE) large language
model developed by Neura Tech AI.

This model is built upon the following base models:

- Neura-Tech-AI/Neuron-4B-Instruct
  Copyright (c) 2026 Neura Tech AI.

- Qwen/Qwen3-4B-Instruct-2507
  Copyright (c) 2024 Alibaba Cloud and the Qwen Team.

- Qwen/Qwen3-4B-Thinking-2507
  Copyright (c) 2024 Alibaba Cloud and the Qwen Team.

Neuron-4B-Instruct is based on the Qwen3 architecture and includes additional
modifications and improvements developed by Neura Tech AI.

The original Qwen3 models are licensed under the Apache License,
Version 2.0. A copy of the Apache License is included in the LICENSE file.

Neuron-6x4B-Instruct introduces additional modifications made by
Neura Tech AI, including but not limited to:

- Mixture of Experts (MoE) architecture
- Six-expert sparse routing design
- Dynamic expert routing configuration
- Expert composition and parameter merging
- Instruction tuning improvements
- Identity customization
- Chat template customization
- Alignment improvements
- Reasoning enhancements
- Multilingual capability improvements
- Coding and software engineering optimization
- Tool-calling optimization
- Long-context support optimization
- Dataset improvements
- Branding and documentation
- Model packaging and distribution

These modifications are Copyright (c) 2026
Neura Tech AI.

Developer

Neura Tech AI
Official AI Research & Development Organization

Project

Neuron

Model

Neuron-6x4B-Instruct

Architecture

Sparse Transformer Decoder
Mixture of Experts (MoE)
Qwen3 Architecture

Base Models

- Neura-Tech-AI/Neuron-4B-Instruct
- Qwen/Qwen3-4B-Instruct-2507
- Qwen/Qwen3-4B-Thinking-2507

Model Highlights

- Approximately 24 Billion Total Parameters
- Six Specialized Experts
- Dynamic Sparse Expert Routing
- Instruction-Tuned
- Multilingual Language Model
- Optimized for Reasoning, Coding, Mathematics, Tool Calling,
  Agentic Workflows, and Long-Context Understanding

Acknowledgment

We sincerely thank Alibaba Cloud and the Qwen Team for releasing the
Qwen3 model family under the Apache License, Version 2.0. Their work
served as the architectural foundation that enabled the development of
Neuron-4B-Instruct and, subsequently, Neuron-6x4B-Instruct.

This NOTICE file is provided solely for attribution purposes and does
not modify, replace, or supersede the terms of the Apache License,
Version 2.0.