[UX]: Explain why onboarding requires selecting a specific LLM model

[simon-ami]

Problem

The “Set up your LLM” step requires a specific model without explaining whether it is a default or how it can be changed, making the choice appear permanent.

Acceptance criteria

• The onboarding copy accurately distinguishes the default model from any global or per-conversation model controls.

[enyst]

Sorry, why does it seem to require a specific model? I don’t think it should- at all.

[simon-ami]

I don't mean that the model is technically permanent. The issue is that during onboarding, it looks like I'm already choosing the model my OpenHands agent will use, without any indication that this is just an initial/default choice.

That makes the decision feel unnecessarily consequential for a new user, especially since the LLM profile/model can later be switched live within a conversation. Something like “Default model (can be changed later/per conversation)” would make this much clearer.