Quickstart
Install B3IQ, claim a local node, create a local API key, and make your first OpenAI-compatible local request.
This quickstart gets one machine from install to a local OpenAI-compatible chat request. Use it when you want local inference working before you think about managed enrollment, earning, or hosted gateway access.
The host agent manages local runtime state and can expose local inference APIs. Install it only on hardware you are allowed to administer.
Install
bashcurl -fsSL https://b3iq.org/install/linux | sudo bash
bashcurl -fsSL https://b3iq.org/install/macos | bash
powershellirm https://b3iq.org/install/windows.ps1 | iex
The installer registers the host service and points you at the local setup
wizard. On Linux, the default listener binds setup and local inference to
127.0.0.1:8831. LAN exposure requires explicit operator opt-in.
Claim The Node
Open the setup wizard
Open the local setup URL printed by the installer. If you need a fresh tokenized URL from the machine, run:
bashb3iq-host wizard url --with-token
Review inventory
The wizard reads hardware, OS, runtime, model, and network posture from the host agent. Browser GPU checks are advisory only; the host agent is the source of truth.
Choose mode
Start with Local Mode unless you are intentionally enrolling the machine into managed or network workflows. Enrolling a new machine into managed mode needs an onboarding code from the B3IQ team. A machine that has already been onboarded instead links to your account by redeeming a claim code at /dashboard/claim.
Create local owner auth
Set the local owner account and create a scoped b3iq_local_ API key for
local inference clients.
Install or select a model
Pick a model profile the machine can actually serve. Larger network models require stronger RAM, VRAM, runtime, and benchmark posture.
Make A Local Request
Store keys in files instead of shell history or process argv.
bashumask 077mkdir -p .secrets# Paste the local key, then press Ctrl-D.cat > .secrets/b3iq-local-key
Call the local OpenAI-compatible chat endpoint. Use a model you actually
installed — list what your node can serve with
curl http://127.0.0.1:8831/v1/models — the model below is only a placeholder:
bash: "${B3IQ_LOCAL_KEY_FILE:=.secrets/b3iq-local-key}"B3IQ_LOCAL_BASE_URL="${B3IQ_LOCAL_BASE_URL:-http://127.0.0.1:8831}"cfg="$(mktemp)"chmod 600 "$cfg"trap 'rm -f "$cfg"' EXIT{ printf 'url = "%s/v1/chat/completions"\n' "${B3IQ_LOCAL_BASE_URL%/}" printf 'request = "POST"\n' printf 'header = "Authorization: Bearer %s"\n' "$(sed -n '1p' "$B3IQ_LOCAL_KEY_FILE")" printf 'header = "Content-Type: application/json"\n'} > "$cfg"curl --fail --silent --show-error --config "$cfg" --data-binary @- <<'JSON'{ "model": "qwen3-8b", "messages": [{ "role": "user", "content": "Run a local B3IQ smoke reply." }], "stream": false}JSON
Local keys are not hosted gateway keys. Hosted calls use scoped
b3iq_gateway_ credentials and route to the machines on your account. See
Hosted Gateway.