Playground

The playground builds a request from the selected model, system prompt, temperature, and max output tokens. Sending it calls the local API. The API checks that the model exists, then refuses the completion because provider routing is disabled.

The response area reports that platform error. It does not invent a model reply, token count, or stream.