or
One account does both: an API key to run models, and a join code to share your GPU and earn.
Your API key · shown only once, copy it now
Dashboard
Run models · API access
Point any OpenAI SDK at this base URL with your API key.
The full key was shown once at signup. Keep it safe.
Share your GPU · join code
Put this code in your node's config. Every GPU that presents it earns for this account.
Node setup
⬇ Download node for Windows (3 MB)
Unzip it, replace its node.toml with this one (your join code is already filled in), then double-click aipool-tray.exe:
The node needs a local OpenAI-compatible model server (LM Studio, Ollama, or llama-server). It only dials out: no port forwarding, and the pool can never run arbitrary code on your machine.
Cluster mode · one-time firewall allow
To let this machine pool VRAM with other GPUs on the same network (giant split models), run this once in an Admin PowerShell. It only allows your own network (LocalSubnet) — nothing from the internet.
New-NetFirewallRule -DisplayName "brAIngrid cluster RPC" -Direction Inbound -Protocol TCP -LocalPort 50050-50060 -Action Allow -Profile Any -RemoteAddress LocalSubnet
Security · change password
Changing your password signs you out on every other device and browser. This one stays signed in.
A passkey signs you in with Windows Hello / fingerprint / face — no password typed.
Your GPUs
No nodes yet. Set one up above; it appears here the moment it connects.
| Node | GPU | Status | tok/s | Jobs | Tokens | Utilized | Est. earned |
|---|
Tokens are metered per job into an audit ledger. Earnings shown are estimates at the Medium token rate until billing goes live. Speed comes from the node's self-benchmark and drives scheduling.