The Panel

Account, keys & quota

Where your credentials live, which key goes where, and how the monthly allowance is actually counted.

Three menus in the Account group handle everything outside the characters themselves: Settings holds the keys to your AI suppliers, API Keys issues the credentials your own software uses, and Plan & Quota shows what you are consuming.

The two kinds of key, once and clearly.
Settings → API Keys = your accounts with suppliers (OpenAI, ElevenLabs, Google, Anthropic, MiniMax). IAMX uses these to generate speech and answers on your behalf.
Account → API Keys = IAMX credentials for your software. They begin with iamx_ and are what your Unreal project or website presents when it connects.

Settings — your provider keys

Settings page with profile, the provider API key list, appearance theme buttons and danger zone
Settings. Each provider row shows whether a key is stored, and lets you replace or delete it.

Keys saved here become the default for every character in the workspace. You paste each one once. A character can still override the voice key on its own Language & Voice tab when you want separate billing for that character — the character's own key always wins.

ProviderUsed forThe key looks like
OpenAILanguage models, speech-to-text, text-to-speech. One key covers all three.sk-…
ElevenLabsPremium voices and voice cloning.sk_… — exactly 51 characters
Google AIGemini and Gemini Live.AIza…
AnthropicClaude models.sk-ant-…
MiniMaxAlternative text-to-speech.key:groupid
ElevenLabs shows you two different values — only one of them is the key. The list in your ElevenLabs dashboard displays a 64-character key ID, which is not a credential and will be rejected. The real key is revealed once, at the moment the key is created or rotated, and it starts with sk_. If you have lost it, open the key's menu and choose Rotate, then copy the value immediately.

Keys are never sent back to your browser after they are stored — the page can only tell you whether one exists. To replace a key, type the new one and save; to remove it, use the button on the row.

Only the workspace owner or an admin can change these keys. Other members of the workspace can use them but not see or replace them.

API Keys — credentials for your own software

API Keys page listing issued keys with their prefix, creation date and last used date
API Keys. Each row shows only the prefix — the full key is displayed once, when you create it.

An IAMX API key is what your Unreal project, your website's embed, or your own application presents when it opens a conversation. Create one per deployment so you can revoke a single device without disturbing the others.

  1. Press New key and give it a name that says where it will live — lobby-kiosk-1, company-website, dev-laptop. The name is for you; it has no effect on what the key can do.
  2. Copy the key immediately. The full value is shown exactly once, in the amber panel. IAMX stores only a fingerprint of it, so nobody — including support — can recover it later. If you lose it, delete the key and issue a new one.
  3. Paste it where it belongs. In Unreal: the IAMX component's CloudAuthToken field. On the web: the ?token= parameter of the WebSocket URL, or the apiKey parameter of the iframe.
The short prefix shown in the list is only there so you can tell two keys apart. It is not a working credential and will not authenticate anything.

Revoking: the trash icon disables a key immediately and permanently. Anything still using it stops connecting on its next attempt. When you rotate a key in production, create the new one and deploy it first, then delete the old one.

Plan & Quota

Plan and Quota page showing current plan with usage bars and the four plan cards
Plan & Quota. The bars turn amber once you pass 80% of an allowance.

How the monthly allowance is measured

IAMX does not charge you per word or per second. Your plan carries a monthly allowance, and every conversation turn draws from it. What matters is that different models draw at very different rates, because they cost very different amounts to run.

What the character is runningRoughly how much of the allowance one turn uses
A fast, small model (mini / flash / haiku class) with standard voice — the default setup — this is the baseline
A large flagship text model (GPT-5.4, Claude Sonnet, Gemini Pro, Grok 4)
The very largest models (GPT-5.5, GPT-5.6 Sol, Claude Opus)5.5×
Gemini Live real-time voice
OpenAI Realtime voice (mini) to 22× (flagship)
A model running locally on your own server0.25×
Sending a camera frame with the turn+1× on top
Real-time voice is the expensive one. A character on OpenAI Realtime can consume in a hundred turns what a normal character consumes in two thousand. It buys you sub-second, interruption-friendly speech — genuinely better for a busy kiosk — but choose it deliberately, and keep an eye on the bar for the first few days.

What happens when the allowance runs out

Conversations stop, with an explanatory message rather than silence. Three different limits can trigger it:

The plans

FreeStarterProEnterprise
Characters1310999
Monthly allowance503,25011,00050,000
Knowledge Bank20 MB100 MB10 GB
Narrative, external APIs, Gemini Live
LiveFace 2D, FaceID, Mindview, voice cloning, OpenAI Realtime
Banking actions, ID verification, phone, on-premise

Features above your plan appear in the character editor with a padlock; clicking one brings you here rather than opening the tab. Changing plan is limited to the workspace owner or an admin.

Buying in Türkiye? Cards issued by Turkish banks can only be charged in Turkish lira, so switch the currency toggle to ₺ TRY before checkout. Everywhere else, leave it on USD.

Where to go next

Watch your usage

Sessions & analytics →