Assess if a more advanced LLM creates better complex prototypes

### What feature do you want to see added?

Try out other LLMs in Azure and assess if they produce higher-quality responses.

### Why do you want this feature? Does it solve a problem?

We have been using GPT-4o-mini however complex prototypes do not always align perfectly to the schema's instructions or user's requests. As prototypes are now more complex, we should assess if a more advanced LLM would be appropriate.

### How should the feature be implemented?

_No response_

### Additional context (if applicable)

_No response_

No files, tests, or entry points are identified in the issue. Start by locating the Azure LLM configuration and prototype-generation entry point, then compare GPT-4o-mini with another Azure model on complex prototypes. Done means a documented assessment of response quality against schema instructions and user requests.

Written by the indexing model from the issue text.