Model Settings
Updated: June 2, 2026
Open the workspace switcher in the sidebar, choose Manage workspace, then select Model to choose the workspace text interviewer model.
This setting applies to text interview conversations in the workspace. It does not replace every model Perspective uses for design, analysis, trust scoring, transcription, voice, or image generation.
Who Can Change It
Workspace Owners and Admins can change the text interviewer model. Members can view the page but cannot save changes.
Choose a Model
The dropdown includes:
- System default - currently the default interview model configured by Perspective.
- A curated list of supported workspace text interviewer models, including GPT-5.5 and Gemini 3.5 Flash when available for your workspace.
Save changes to apply the selected model to future text interviewer turns. If the saved model is no longer supported, Perspective falls back to the system default until an Owner or Admin chooses a supported model.
Choose GPT-5.5 when you want stronger reasoning over long text conversations. Choose Gemini 3.5 Flash when you want fast text replies with a large context window. Most teams should stay on System default unless they are intentionally comparing model behavior.
How Model Selection Works
For normal text conversations, Perspective resolves the model in this order:
- A request-level model override, when one is supplied by an internal/debug flow.
- The workspace text interviewer model from Settings.
- The system default interview model.
This setting controls participant-facing text interviewer turns. The model menu inside the design chat can offer additional design-assistant models, but that menu does not change the workspace text interviewer default.