API & Node
Documentation

1. Run the node (no install needed)

npx neuronix-node

This downloads and runs the Neuronix node automatically. No cloning, no setup.

2. Login with your Neuronix account

The node will prompt you for your email and password on first run. Create an account at neuronix-nu.vercel.app/signupif you haven't already.

3. Done — you're earning

The node auto-detects your hardware, downloads the best AI model for your GPU, and connects to the network. Leave it running and you'll earn for every task processed.

Alternative: Global install

If you want to run it without npx every time:

npm install -g neuronix-node
neuronix-node

Skip the login prompt (optional)

Set environment variables to authenticate automatically:

export NEURONIX_EMAIL="you@company.com"
export NEURONIX_PASSWORD="your-password"
npx neuronix-node

"Registration failed" on startup

Check your internet connection. If the problem persists, verify you have an account at neuronix-nu.vercel.app/signup and that your email/password are correct.

"Model setup failed" error

The model download may have been interrupted. Delete the partial file from ~/.neuronix/models/ and restart the node. It will re-download automatically.

Node shows 'online' but no tasks are coming in

This is normal when the network has more providers than tasks. Tasks are dispatched based on priority and VRAM — higher-end GPUs get tasks first. Leave the node running and tasks will come.

High CPU/GPU usage

This only happens during active task processing (a few seconds at a time). Between tasks, the node is idle. If usage stays high, check that no other process is using your GPU.

How do I update the node?

The node checks for updates on startup. To update manually, run 'npx neuronix-node@latest' or if globally installed, 'npm update -g neuronix-node'. Your config and models are preserved.

Can I run multiple nodes on one machine?

Not recommended — each node binds to the same GPU. Run one node per machine for best results.

import requests

API_KEY = "az_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx"
BASE_URL = "https://api.neuronix.io/v1"

def run_inference(prompt: str, model: str = "llama3-8b") -> dict:
    response = requests.post(
        f"{BASE_URL}/inference",
        headers={
            "Authorization": f"Bearer {API_KEY}",
            "Content-Type": "application/json",
        },
        json={
            "model": model,
            "prompt": prompt,
            "max_tokens": 1024,
            "temperature": 0.7,
        },
    )
    response.raise_for_status()
    return response.json()

result = run_inference("Summarize the risks of a Class B office investment.")
print(result["result"]["text"])
print(f"Cost: ${result['cost_usd']:.6f}")
print(f"Tokens: {result['result']['usage']['total_tokens']}")

const API_KEY = "az_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx";
const BASE_URL = "https://api.neuronix.io/v1";

async function runInference(prompt: string, model = "llama3-8b") {
  const res = await fetch(`${BASE_URL}/inference`, {
    method: "POST",
    headers: {
      Authorization: `Bearer ${API_KEY}`,
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      model,
      prompt,
      max_tokens: 1024,
      temperature: 0.7,
    }),
  });

  if (!res.ok) {
    throw new Error(`Neuronix API error: ${res.status}`);
  }

  return res.json();
}

const result = await runInference(
  "Write a property listing for a 4-bedroom colonial in Westport, CT"
);

console.log(result.result.text);
console.log(`Cost: $${result.cost_usd}`);
console.log(`Tokens: ${result.result.usage.total_tokens}`);

curl -X POST https://api.neuronix.io/v1/inference \
  -H "Authorization: Bearer az_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "llama3-8b",
    "prompt": "Analyze Q1 2026 multifamily market trends in Phoenix, AZ",
    "max_tokens": 1024,
    "temperature": 0.7
  }'

# Add "stream": true to receive Server-Sent Events
curl -X POST https://api.neuronix.io/v1/inference \
  -H "Authorization: Bearer az_live_xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  --no-buffer \
  -d '{
    "model": "llama3-8b",
    "prompt": "Write a detailed market analysis for Austin, TX",
    "max_tokens": 2048,
    "stream": true
  }'

# Each SSE event:
# data: {"delta": "text chunk", "finish_reason": null}
# data: {"delta": "", "finish_reason": "stop", "usage": {...}}