Quick start

1. Sign in

Open Janus. With SSO configured, choose “Continue with corporate SSO” (or “Continue with” your identity provider): your account is created automatically on first sign-in, and your group memberships are read from the identity token each time. Without SSO, sign in with the Janus account your administrator created for you (you will be asked to choose your own password the first time).

On a brand-new gateway without SSO, the first visit asks you to create the administrator account; with SSO, administrators come from the configured admin emails or admin groups. If every licensed seat is in use, a new user’s first sign-in is refused until an administrator adds seats.

You do not need to be added to an allowlist to sign in. What you can do is controlled by model grants.

2. Create a token

Go to API Tokens and choose “Create token”. Name it after the tool that will use it — that is what makes revoking a single integration safe later. Pick its Token context — Personal or one of your teams — which decides the models and quotas the token uses.

The value is displayed once. Janus stores only a SHA-256 digest and genuinely cannot show it again.

3. Point your tool at the gateway

Janus speaks the OpenAI API. Set the base URL to your gateway’s API root, ending in /v1 (Help shows the exact value), and use your token as the API key. Everything else in your client stays the same.

export JANUS_BASE_URL="https://your-gateway"
export JANUS_API_KEY="janus_…"

curl "$JANUS_BASE_URL/v1/chat/completions" \
  -H "Authorization: Bearer $JANUS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "MODEL_NAME",
    "messages": [{"role": "user", "content": "Hello from Janus"}]
  }'

4. Watch it land

Open your dashboard. The request appears within a few seconds with its token counts, latency, and attributed cost.

If the call was refused, the error carries a stable code — look it up in the error catalog.