json output for llm models --options and text verbosity.

[grota]

Hi @simonw
I would like to work on two improvements :

• json output for llm models --options
• text verbosity

The first in this main llm project, the second in your llm-openai-plugin since it's a responses-only parameter and for models that actually support it (gpt-5 family for now).

Is this something that you would actually be interested in having?