NEXUS Docs

NEXUS is a decentralized AI network. Instead of renting cloud GPUs, prompts are served by a network of independent workers who lend their GPUs and earn $NEXUS per job. You pre-pay a small $NEXUS credit, and every message is billed instantly from that balance — no per-prompt signature, no subscription.

• Pay as you go. Deposit $NEXUS once; each prompt deducts a flat per-model fee.
• Open-weight models. Served on real worker GPUs (Qwen, Gemma, Llama, and more).
• Private. Prompts aren't stored, and workers never see who you are.
• Developer-ready. An OpenAI-compatible API drops into any existing app.

1. 1

   Deposit $NEXUS

   Sign one transaction to top up your credit balance. The $NEXUS lands in the treasury and your off-chain credit is updated.

2. 2

   Prompt

   Send a message from the chat or the API. The gateway deducts the model's fee from your credit before dispatching the job.

3. 3

   A worker runs it

   The coordinator routes the job to an online worker that serves your chosen model and streams tokens back to you in real time.

4. 4

   The worker gets paid

   On completion, the worker earns its share of the fee in $NEXUS; the rest accrues to the protocol treasury.

The fastest way to try the network:

1. 1. Open the Chat and sign in with your wallet.
2. 2. Click Deposit and top up some $NEXUS credit.
3. 3. Pick a model, type a message, and hit send.

Building an app instead? Create a key on the Developers page and jump to the Developer API.

Pricing is a flat fee per message, set per model and charged in $NEXUS. Pass the modelidin chat or API requests.

The API is OpenAI-compatible. Point any OpenAI SDK at https://nexus-production-ad34.up.railway.app/v1, authenticate with your nex_live_key, and each request is served by the network and billed to that key's wallet.

TypeScript · OpenAI SDK

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://nexus-production-ad34.up.railway.app/v1",
  apiKey: process.env.NEXUS_API_KEY, // nex_live_…
});

const res = await client.chat.completions.create({
  model: "qwen3-32b",
  messages: [{ role: "user", content: "Hello from my app!" }],
});

console.log(res.choices[0].message.content);

curl

curl https://nexus-production-ad34.up.railway.app/v1/chat/completions \
  -H "Authorization: Bearer nex_live_…" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3-32b",
    "messages": [{ "role": "user", "content": "Hello!" }]
  }'

Streaming (stream: true),/v1/models, and the standard usage fields are all supported. Manage keys on the Developers page.

Lend your GPU and earn $NEXUS for every prompt you serve. A worker connects to the gateway over WebSocket, registers its hardware, and waits for jobs.

Start a worker

cd worker
npm install

# Serve a real model with Ollama (recommended):
#   OLLAMA_URL=http://localhost:11434
#   ollama pull qwen3:32b
# Set your payout wallet:
#   WORKER_WALLET=<your-solana-address>

npm run start

Without an OLLAMA_URL the worker runs in mock mode (handy for testing). Earnings accrue per job and can be claimed to your wallet. Track earnings and network stats on the Earn page.