Help with deployed instances accessing the inference models with API_KEYS

I see the following error when following the guide - https://huggingface.co/learn/mcp-course/unit2/gradio-client

smolagents.utils.AgentGenerationError: Error in generating model output:
You must provide an api_key to work with together API or log in with huggingface-cli login.

I had created a PR to address some more small things, was planning to add it there, if someone can figure this part out.
PR - https://github.com/huggingface/mcp-course/pull/78

Start by following the Gradio client guide at the linked Hugging Face course page and reproduce the API key error for a deployed instance. Then inspect PR #78 for the related documentation changes. Done means the guide explains authentication clearly and deployed instances can access the inference models without this error.

Written by the indexing model from the issue text.