Settings
Global settings (Settings → ...)
Global settings apply across all your workspaces.
Model Library
Manage your AI provider API keys and the list of available models. See Model Library.
Operator Models
Configure the AI models used to interpret coding agent terminal output. See Operator Models.
Profile
Update your display name and account details.
Workspace settings
Each workspace has its own settings, accessible from the workspace header → Settings.
AI provider
Change the AI model used in this workspace. You can switch models without losing your conversation history or workspace files. The new model takes effect on the next prompt.
| Field | Description |
|---|---|
| Provider | Anthropic, OpenAI, Google, Groq, or Together |
| Model | The specific model (from your Model Library) |
| API key override | Use a different key for this workspace only. Leave blank to use your global key |
| Agent type | Interactive or one-off, and which agent CLI to use |
Repository
Read-only information about the workspace's GitHub repository:
| Field | Description |
|---|---|
| GitHub repo | The repository this workspace is connected to |
| Mode | Fork (isolated copy) or Attach (shared repo) |
Sandbox resources
If sandboxes are enabled on your deployment, you can adjust the CPU and memory allocated to this workspace's microVM. Changes take effect the next time the workspace is opened.
Server configuration
Optional deployment target settings (server URL, Docker registry, environment). Used when the workspace is configured to deploy artifacts to a remote server.
Changing models mid-session
You can switch the AI model while a workspace is open — you do not need to close or recreate the workspace. The model switcher is in the workspace header. The next prompt you send will use the new model. Conversation history is preserved.