A dialogue allowing a user to choose between "model A", "model B" and "model C"


Let people take control of the model they want to use. This means a previous model could continue providing a service, rather than the ‘latest’ model.

For example, someone could choose a model that has been trained to perform in a specific way, that the user wants to preserve or return to, or in professional settings where reliability of output is important.

IF thinks this pattern helps to engender a trustworthy relationship between the person and the AI system, by giving people choice of model.


  • Increases user’s agency and control.
  • Enables a more meaningful and active relationship between the user and the system itself.
  • Supports a user through a change in their life, for example a new job.


  • The UI must be clear and unambiguous so it is not misunderstood by the user (or the AI).


  • Claude →

    Anthropic gives controls for Claude Pro users to improve their experience. For example, users can select which model version of Claude they want to power their chat experience.