Chat Settings
Choose the model Forge runs on, how it talks to you, and how often it asks for approval.
Three settings sit below the chat input: Model, Audience, and Approval mode. When the sidebar is narrow they collapse into a single Settings button.
Model
By default Forge runs on the model configured for your platform — the button reads Default model, and there's nothing to set up.
If you'd rather use your own model, click the button and fill in the Custom model form. Any OpenAI-compatible endpoint works:
| Field | Example | Notes |
|---|---|---|
| Base URL | https://api.openai.com/v1 | Your provider's or company gateway's endpoint |
| API key | sk-... | Kept in your browser only |
| Model | gpt-4o | Model name as your provider expects it |
Click Save and the change applies immediately — no reload needed, and it covers all your conversations in this browser. Reset to default switches back to the platform model.
Your API key is stored only in this browser and sent along with each request — the server never saves it. Clearing your browser data removes it.
When is a custom model useful?
- Your company routes LLM traffic through its own gateway.
- You want to compare how Forge performs on a different model.
- Your platform's default model has usage limits you'd rather not share.
Audience
The Audience setting controls how Forge explains things — not what it builds. The same request produces the same workflow; only the narration changes.
| Level | How Forge talks |
|---|---|
| For beginners | Plain language, no jargon — explains what workflows and connectors are along the way |
| For developers (default) | Developer terms are fine, but Studio specifics still get explained |
| For experts | Terse, full jargon — assumes you know the platform |
Pick the level that matches you, not the task. If Forge's answers feel either patronizing or too dense, this is the knob to turn.
Approval mode
Controls how often Forge pauses for your confirmation before changing something — from asking about everything to running fully autonomously. Unlike the other two settings, it applies per conversation.
The modes and their trade-offs are covered in Approvals.