Skip to main content

Deploying a Model

Deploying a model means making it callable under a specific namespace endpoint. Until a model is deployed to at least one namespace, it is not addressable by API keys scoped to that namespace.

Provider models are deployed by your CleverThis administrator and are typically available workspace-wide without additional configuration. Actor endpoints, on the other hand, must be explicitly deployed to a namespace before they are callable — this is covered in detail in Creating an Actor.

Deploy via the UI​

  1. Navigate to Library → Models and open the model you want to deploy.
  2. Click Deploy to Namespace.
  3. Select the target namespace from the dropdown. If you have not created a namespace yet, see Creating a Namespace.
  4. Optionally assign an endpoint alias — a short name that becomes the model ID callers use within this namespace (e.g. fast-chat instead of openai/gpt-4o-mini).
  5. Click Deploy. The deployment appears in the Deployments list with status running.

Deploy via the API​

You can also deploy models programmatically using the management API (ct_mgmt_ key). See the A2A control plane → Namespaces reference for the exact method calls.

Verifying the deployment​

After deploying, call GET /v1/models with an inference key scoped to that namespace. The newly deployed model should appear in the response list.

Coming soon

Step-by-step screenshots and documentation for per-endpoint rate limit and cost cap configuration are being added to this page.