Create virtual model config
Create a virtual model config (VMC)
Query parameters
The Team identifier to perform the request on behalf of.
The Team slug to perform the request on behalf of.
Response
Team (owner) that owns this VMC.
Client-facing alias used as the model slug in Gateway calls.
Human-readable name for UI.
Optional description for UI.
Whether this VMC is soft-deleted.
UI lifecycle status: draft, active, or archived.
Visibility in listings: public, internal, or stealth.
User id that last updated this VMC.
VMC kind: alias, relay, or router.
For kind=relay: URL the gateway forwards requests to as a transparent proxy.
The concrete model-provider instance this VMC resolves to.
Ordered list of providers to try as fallbacks on failure.
Restrict routing to only these providers.
Arbitrary per-provider AI SDK options, keyed by gateway provider slug.
Canonical model slug this VMC maps to (e.g. "creator/model"). Not used by kind=router.
For kind=router: ordered candidates, model slugs or router references. Otherwise: fallback models.
For kind=router: how to order candidates.
For kind=router: capability tags a candidate must have.
BYOK credential IDs allowed for this VMC.
Observability tags attached to requests through this VMC.
Rank eligible providers by an attribute.
Limit providers to those with these features.
Use caching if available.
Service tier for providers that support it.
Only use providers with zero data retention.
Only use HIPAA-compliant providers.
Only use providers that will not train on your prompts.
Only use fastest providers with short timeouts.
Allow fallback from fast to standard providers on failure.
Creation timestamp (epoch ms).
Last update timestamp (epoch ms).