Configure Default Fallback

Use your deployed API Router Gateway for model requests. Check the deployed version and required permissions. Example addresses are placeholders.

Default Fallback is a cross-model recovery policy. When all available providers for a directly requested model fail, Smart Studio can retry with the configured fallback model.

Before You Start

• Choose a fallback model that accepts the same request modality and format as the primary model.
• Confirm that the fallback has sufficient availability and an acceptable cost profile.
• Record the current policy and prepare a controlled verification request.

Configure the Fallback

1. Open Account → Settings (in your deployed Console).
2. In Default Fallback Model, enable the policy.
3. Search for and select the backup model.
4. Select Save.

The fallback model is a recovery path, not a routine model-selection strategy. Use Auto Routing when the application intentionally delegates model choice.

Auto Routing and Fallback

	Auto Routing	Default Fallback
Trigger	The request sends model="auto"	All providers for a directly requested model fail
Model choice	Router selects from the candidate pool	Tenant-configured backup model
Primary goal	Dynamic model selection	Cross-model recovery

Requests using model="auto" use the Auto Routing candidate pool and do not rely on the Tenant Default Fallback policy.

Verify the Result

1. Confirm that the policy saves with the intended fallback model.
2. Use an approved failure test or staging environment; do not deliberately disrupt production providers.
3. Inspect History to confirm whether the fallback handled the request.
4. Review Usage for the resulting model consumption.

Troubleshooting

• The policy cannot be enabled without selecting a fallback model.
• A fallback cannot correct an invalid request format or an authentication failure.
• If the primary and fallback models accept different modalities, the retry can still fail.
• Use trace information and History to distinguish provider failover from cross-model fallback.

Next Step

Review Provider Selection for provider-level routing, then monitor History and Usage after enabling fallback.