AI Models
Which models you see in the picker, which one is your default, and the temperature slider.
Your organisation decides which models exist. This section decides which of them you see, and how they behave for you.
Hiding models you never use
Every model your organisation offers is listed. Hide any you do not want cluttering the picker in the message box, Show to bring one back.
Hiding is cosmetic and reversible. It does not remove your access, and it does not stop Asteria choosing a suitable model for background work.
Worth doing if your organisation offers a dozen models and you use two. A shorter picker is a faster decision.
Your default model
Set as default marks the one every new conversation starts with.
Leave it on whatever your administrator chose unless you have a reason. A better brief on a smaller model beats a vague one on a larger model nearly every time, and the reasons to change are shorter than you would expect. They are listed in The message box.
Refresh model list picks up models your administrator has just added, if a new one has not appeared yet.
Response temperature
A slider from precise to creative, controlling how varied the answers are.
Low (0 to 0.3) for anything factual: figures, policy questions, code, summarising a document. The same question gives close to the same answer each time, which is what you want when the answer has to be right.
High (0.7 to 1.0) for brainstorming, naming, drafting alternatives. More variety, less predictability.
In between for everyday use, which is where the default sits.
Most people should leave this alone. It is a genuine trade-off rather than a quality dial: turning it up does not make Asteria cleverer, it makes it less repeatable. If answers feel bland, the fix is usually a better brief, not a higher temperature.
This is your personal setting. An agent can override it for its own runs, and a workflow step can set its own.