N NODIFYAI
ONE GATEWAY · SIX MODELS · NO SETUP

Every model you use.
One calm place to use it.

Nodify AI gives you a fast chat workspace and an OpenAI-compatible API in one place. Compare the models, see exactly what things cost, then create an account when you’re ready.

Download desktop app
No credit card to start · 300K weighted tokens/month free · 50K per 5 hours
6 live models SSE streaming BYOK ready 0 overage bills
THE MODEL LINEUP

Pick the right tool for the job.

Six hosted models behind one API. Switch in the model picker or change a single field in your request.

Loading models…
HOW IT WORKS

From zero to running in three steps.

No infrastructure to manage. Workers connect outbound, so nothing on your network needs a public IP.

01

Create your AI account

Email, password, and a six-digit code. This account is separate from your normal Nodify host account.

02

Chat or call the API

Use the web console for thinking out loud, or generate an API key and drop it into any OpenAI client.

03

Track it in one place

See usage, keys, preferences, contributor credits, and live cluster health from the same dashboard.

BUILT FOR SPEED

Quietly fast. Predictably priced.

Low-latency streaming

SSE goes straight from the worker to your browser or client, with buffered proxies disabled.

Fair, predictable caps

Every account has a monthly weighted-token allowance and a 50K/5-hour rolling cap. Paid plans automatically spill over to Free.

Contributor credits

Run a GPU worker and earn credits for the tokens you serve. Enough credits unlocks Pro automatically.

Bring your own key

Point a model at your own provider from Preferences. Your key stays in your account settings.

Saved conversations

Every chat is stored to your AI account, searchable from the sidebar and resumable later.

OpenAI-compatible

Use the SDKs, CLIs, and tools you already know. Only the base URL and key change.

PRICING

Simple plans. No overage surprises.

Usage is counted in weighted tokens. A 1x model spends 1 token per token; a 2x model spends 2, and the 4x reasoning model spends 4. Prompt and completion tokens both count. Free includes 300K/month with a 50K/5-hour rolling window, and paid plans keep going on Free when exhausted.

Loading plans…
Weighted example: 100K tokens on a 2x model uses 200K of your allowance. What counts: the prompt you send plus the reply you receive. Resets: monthly allowance every 30 days; the 50K rolling window every 5 hours.
THE API

Drop it into anything.

Generate a key in your dashboard, set your base URL, and call Nodify like any OpenAI-compatible endpoint. Streaming, usage, and model discovery are built in.

  • Base URL: https://ai.nodifyhost.dynv6.net/v1
  • Discover models with GET /v1/models
  • Check your allowance with GET /v1/usage
curl https://ai.nodifyhost.dynv6.net/v1/chat/completions \
  -H "Authorization: Bearer $NODIFY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-3.5-4b",
    "stream": true,
    "messages": [
      { "role": "user", "content": "Hello Nodify" }
    ]
  }'
GOOD QUESTIONS

Before you sign up.

Is this the same account as my normal Nodify host?

No. Nodify AI has its own accounts, sessions, keys, and usage. Your main Nodify panel login does not work here, and nothing AI-related appears on the main site anymore.

Do I need a payment method to try it?

No. Creating an account puts you on the Free plan with 300,000 weighted tokens per month and a 50,000/5-hour rolling cap. Paid checkout is not enabled yet, but you can create an account now and keep your settings.

What happens when I run out of tokens?

Free requests pause with a clear insufficient_quota error until the 5-hour or 30-day window refreshes. Paid plans automatically continue on the Free allowance instead of stopping. Failed requests do not bill you, and there are no silent overage charges.

Can I use my own API key?

Yes. Add a base URL and key under Preferences, toggle BYOK, and inference will go to your provider instead of the Nodify cluster. Your allowance is not consumed on BYOK requests.

Which OpenAI clients work?

Anything that accepts a custom base URL and bearer key: the official SDKs, LangChain, Open WebUI, Continue, Cursor, and most CLI tools. Set the base URL to the API URL shown above.

Where do the models actually run?

On contributor GPU workers that connect outbound over WSS. The gateway never needs to reach them directly, so home and NAT machines work without port forwarding. No worker IPs or identities are shown in the dashboard.

Can I contribute my own GPU?

Yes. Download the Nodify AI desktop app and search for “Nodify AI” in Windows after installing. On first launch it opens your browser to connect your AI account, then finishes setup automatically: llama-server.exe is included in the download, your GPU backend and models folder are detected, and a starter model is cached in the background. It sits in the tray, shows your usage and 5-hour window, and contributes spare GPU capacity by default; you can switch on idle-only mode in Settings if you prefer.

READY WHEN YOU ARE

Start free. Stay because it’s simple.

Read everything above, then create an account in under a minute.