Zero open ports · one agent

Secure access to your AI machines from anywhere.

A small agent on your GPU box dials out to our relay. Nothing on your network opens up. Chat with your models from your phone, SSH in from a browser, share ComfyUI with one link.

Free for one machine · no credit card
What you do with it

Your machine, from wherever you are.

Ollama from your phone

Pick a model, start typing. Tokens stream from your home GPU over LTE. Works with LM Studio too.

SSH from the browser

Open a full terminal on any of your machines from the dashboard. No keys to copy, no VPN.

Share ComfyUI with one link

Send a link. Your friend sees the gallery without installing anything. Set an expiry or a password, or revoke it any time.

How it works

Three steps. Under a minute.

01

Download the agent

Install it on the machine you want to reach. It runs as one quiet process by the clock.

02

Sign in

The agent opens your browser and shows a code. Confirm it once and the machine is tied to your account.

03

Your services show up

The agent finds what's running. Nothing is public until you create a Share.

Prefer a terminal?
$ curl -fsSL https://anyfyio.com/install.sh | sh
Security

Nothing listens on your network.

Mutual TLS

The agent and the relay both present certificates. Neither side talks to a peer it can't verify.

The certificate is the identity

Each device gets its own certificate when you enroll it. Remove the device and the certificate stops working.

Outbound only

The agent dials out on 443. No inbound ports, no router changes, nothing for a scanner to find.

Per-share API keys

Every Share has its own key. Rotate or revoke one without touching the rest.

OpenAI-compatible

Works with any OpenAI SDK.

Point base_url at your Share. Anything that speaks the OpenAI API now runs on your own GPU, from editors to agents to notebooks.

/v1/chat/completions /v1/models stream: true
from openai import OpenAI
 
client = OpenAI(
    base_url="https://k3f9x2.share.anyfyio.com/v1",
    api_key="afy_sk_7Qm…",  # this Share's key
)
 
r = client.chat.completions.create(
    model="llama3.3:70b",
    messages=[{"role": "user", "content": "Hello from the bus"}],
)
Pricing

Free for one machine.

Compare plans →
Free $0 1 device · 1 GB/mo
Team $29 /mo 20 devices · 200 GB/mo
Business $99 /mo unlimited · 2 TB/mo
FAQ

Straight answers.

Is this a VPN?

No. anyfyio doesn't create a network between your machines. The agent connects out to our relay, and the relay forwards specific requests to specific services on that machine. Nothing else on your network changes.

Do I need to open ports on my router?

No. The agent makes an outbound connection on 443. No inbound ports, no port forwarding, no firewall rules.

What's the difference between a Share and Reach?

A Share is a public link anyone can open, with an optional password, expiry and API key. Reach is private to your org: only signed-in members can open it.

Why not Tailscale?

If you want every device to reach every other device, use a mesh VPN like Tailscale. anyfyio is for reaching a few AI machines from anywhere, with chat, ComfyUI and Jupyter built in, and nothing to install on the phone except the app.

What happens when my machine goes offline?

Your Shares stay put and show a clear offline page. The agent reconnects on its own when the machine comes back, and the dashboard shows when it was last seen.

Can you see my prompts?

Traffic is encrypted between your device and the relay, and the relay forwards it without storing request bodies. Audit logs record who opened what and when, not the content.

Put your first machine online.

Free for one device. Takes about a minute.

Not ready to install yet? Join the waitlist and we'll tell you when new features land.