System Prompt
The system prompt tells the AI how to behave. It defines the chatbot's personality, knowledge boundaries, response style, and any specific instructions.
Key Elements
- Model Selector — Choose which AI model powers your chatbot
- Adjust — Fine-tune the model's response settings
- System Prompt Text — The instructions that define your chatbot's behavior
Model Selection
Click the model name to open the model picker. Models are grouped by who provides them.
Free
Included with every plan and ready to use with no setup:
- GPT-6 Luna — The default for new chatbots.
- Claude Haiku 5.5 — Anthropic's fast model.
With your own API key
Add your own OpenAI or Anthropic key to use these models. In the model picker, click "Use your own keys" to add one; models on your key are listed under "OpenAI · your key" or "Anthropic · your key".
- OpenAI — GPT-6 Luna
- Anthropic — Claude Opus 5.5, Claude Sonnet 5.5, and Claude Haiku 5.5
Team members who can't add keys see "Ask an admin to add more models" instead.
Older models
A chatbot created on a model that has since been retired keeps working on it, and the picker marks that model "Old". Once you switch to a current model, you can't switch back.
Adjust Parameters
Click "Adjust" to open the configuration panel. The settings you see depend on the model:
- Stream — Show the reply as it's written (recommended: Yes). Every model.
- Max Tokens — The longest reply the model may write (default: 2048). Every model.
- Temperature — Fixed at 1 for GPT-6 Luna, so the slider is locked. Claude 5.5 models don't use it, so it's hidden.
- Top P, Frequency Penalty, Presence Penalty — Shown only for models that accept them. Claude 5.5 models don't.