Would be nice to have the option to hide what model a custom agent is using.
There is some bias against certain LLMs and this could help prevent lack of adoption of a helpful tool should it use a model that someone doesn't tend to like.