Choose what powers your conversation.

Your passport, guided discovery and preparation tools work without a model key or subscription. Add optional AI when it helps.

Open my passport studio Sign in

Managed AI, when you need it

Sign in to see your balance in service credits.

AUD 10 per top-up. Optional studio message interpretation costs AUD 0.10 per successfully completed request. Task prices include model processing and service costs; credits are not raw provider tokens.

Manual top-ups, no automatic renewal and no short credit expiry. Failed managed tasks release their reservation. Free guided mode stays available.

I have reviewed the one-time purchase scope, terms and refund information. No automatic renewal.

Bring your provider

Your provider bills you directly. These requests do not spend ExiLnk credits. A ChatGPT/Claude chat subscription is separate from developer API access.

Provider connection detailsConnection nameProviderModel IDHTTPS API base URLAPI protocolDedicated API keyTest this selected provider using synthetic text and store my credential encrypted on the server. The test may incur a small provider charge.Do not enter an organisation master/admin key. A connection test checks a small structured task; feature compatibility varies by model. Custom gateways must have a public HTTPS endpoint. Local models use the bridge below.

Saved connections

No saved cloud connections. You can still use guided mode or pair a local model.

Use a local model

Ollama, LM Studio and compatible local servers can power interpretation without ExiLnk AI credits. Your hardware and electricity still have a cost. A local model does not make your cloud passport offline.

Start your model server on your computer and note its model ID.
From the ExiLnk source folder, run the local bridge using the setup guide. It only listens on your computer and prints a temporary pairing token.
Open the passport studio, choose Local model and enter that token. It stays in page memory and is never sent to ExiLnk.

For developers: npm run ai:local -- --model YOUR_MODEL --origin https://YOUR_EXILNK_SITE. Default model endpoint: http://127.0.0.1:11434/v1. For LM Studio add --endpoint http://127.0.0.1:1234/v1.

Browser local-network permission may be required. Keep the inference server private; do not expose an unauthenticated model port to the internet. Enterprise connections use the approved public gateway form above; private networking needs a separately provisioned deployment.