The questions teams ask before they switch
Before a team moves production traffic, the questions are practical:
- Which models can I call today?
- How much will this cost at my volume?
- If something fails, can my team find the request and fix it?
- Can finance reconcile the spend without building a spreadsheet from scratch?
This post answers those questions in product terms.
1. Models you can actually call
The model catalog shows model IDs, capabilities, context windows, and prices together. The same IDs work in the dashboard, /v1/models, and OpenAI-compatible calls.
That means your team can pick a model from the site, paste the ID into code, and see the same model in usage and billing records later.
2. Errors your app can handle
Your app calls the same SeaLink base URL and model ID. If a request fails, the response uses a consistent OpenAI-compatible error shape and includes a request ID.
Your team can retry, switch model IDs, or send the request ID to support. /docs/error-codes lists the cases worth handling in code.
3. Billing
Top up once through hosted checkout, then spend that balance across models. The dashboard breaks usage down by model, key, and task metadata.
Set monthly caps on API keys to stop runaway usage. Finance can pull top-ups, receipts, and usage records from one place instead of rebuilding spend from raw API logs.
The pricing page shows SeaLink retail rates in USD before you call the model.
How to verify
Start with three checks:
1. Pick a model from /models and copy its ID.
2. Run the Quickstart with your SeaLink API key.
3. Open the dashboard and confirm the request appears in usage.
If anything does not line up, send the request ID to contact@sealink.io.