Chat settings
Tune the system prompt, temperature and context length for each chat.
Chat settings shape how the AI answers. Set the defaults every new chat starts with in Workspace Settings:
- Click the gear next to the workspace switcher at the top of the sidebar.
- Open the Defaults tab.
- Choose the Model, edit the Prompt, and open Advanced Settings for temperature, context length and more.
- Click Save.
The settings
Model
The model that writes the replies. See Choosing a model.
Prompt (system prompt)
Instructions the AI follows for the whole chat. The default is "You are a helpful AI assistant." Replace it to set a role, tone or format, for example:
You are a senior copy editor. Reply with the edited text first,
then a short bulleted list of what you changed and why.
Temperature
How creative or predictable the answers are.
- Lower (towards 0) - focused, consistent answers. Good for facts, code and data.
- Higher - more varied, creative answers. Good for brainstorming and writing.
Context Length
How much of the conversation and attached content the model reads each time it replies. Higher values remember more of a long chat but cost more tokens. The maximum depends on the model.
Chats Include Profile Context
When on, the AI also sees the profile context you wrote in your profile - for example your role or preferred language.
This chat only
To switch the model for just the current chat, use the model selector at the top of the chat.
Assistants have their own settings
When you chat with an assistant, its model, prompt and other settings are used instead.
Was this article helpful?