API Catalog
The API Catalog lists all services on the platform. You see every service, also those you cannot use yet. Open it in the portal under Platform, then API Catalog.
Sections
- Platform models: models that the platform runs for all users.
- My models and adapters: your own model servers and adapters, which you start through the Compute API. A tag shows whether an entry is a model or an adapter. Only the users on the entry's whitelist see it. You have access from the start and do not request it.
- Infrastructure: platform services, for example the Compute API.
Access states
| Badge | Meaning |
|---|---|
| Granted (green) | You have access. Click the row to open the service. |
| Private (green) | Your own model or adapter. You have access from the start. |
| Pending (yellow) | You asked for access. An admin has not reviewed the request yet. |
| Request Access (grey button) | You have no access yet. Click the button to ask for it. |
Services that need no approval, for example the Compute API, show as Granted without a request.
Request access
- Find the service in the catalog.
- Click Request Access in its row. The badge changes to Pending.
- An admin reviews your request. When the admin approves it, the row turns green.
You do not need to contact anyone. The admin sees your request.
Open a service
Click a green row to open the detail page. For a model, the page has three tabs:
- Overview shows the endpoint path, the methods, the model name (or the
adapter and its base model), and an example request with
curl. - Usage shows latency and token throughput.
- Infrastructure shows the health of the model and its load, for example running and waiting requests, the cache usage, and the request volume.
The example request uses the full endpoint URL. If an endpoint path ends in a
wildcard, the example shows ENDPOINT. Replace it with one of the listed
paths. Send your API key in the apikey header.
Filter and search
The buttons All, AI Models and Infrastructure filter the list and show the number of entries. The search field finds services by endpoint path, backend, model name, or base model.