AI Catalog
The AI Catalog (Platform → AI Catalog) lists all models available on the platform. You can see every model regardless of whether you currently have access to it.
Your Private Models and Adapters
When you start a private model server or deploy an adapter through the compute API, it shows in the catalog under My models and adapters, between the platform models and the infrastructure services. A tag says if the entry is a model or an adapter. Only the users on its whitelist see it; other users do not. You do not request access to these entries: you have access from the start. Search and the filter buttons work on them too. Click a row to see its endpoint.
Access States
| Badge | Meaning |
|---|---|
| Granted (green) | You have access. Click the row to see endpoint details and example usage |
| Pending (yellow) | You have submitted a request and it is waiting for admin approval |
| Request Access (button) | You do not have access yet. Click to submit a request |
The Compute API needs no request. It always shows as Granted.
Requesting Access to a Model
- Find the model you want in the catalog
- Click Request Access on its row
- The button changes to Pending immediately
- A platform admin will review your request. Once approved, the model turns green
You do not need to contact anyone manually. The admin will see your request in their panel.
Using a Granted Model
Click on any green row to open the model detail page. It shows:
- The full endpoint URL
- The exact
curlcommand to use - The model name to pass in the request body
Use your API key from Developer → API Keys in the apikey header.
Filtering the Catalog
Use the filter buttons at the top to view:
- All: every model and service
- AI Models: large language models and other ML endpoints
- Infrastructure: other platform services
You can also use the search bar to find a specific model by name or endpoint path.