Loading catalog.
Loading catalog.
general-mini
This table explains how the alias turns a friendly model choice into a governed provider call, including who may use it, what data it may process, and which provider features are exposed.
| Field | Current value |
|---|---|
Alias The friendly name and stable key that callers select. Teams use the alias instead of hard-coding a provider model ID so routing and approvals can change centrally. | Sandbox general mini legacy aliasgeneral-mini |
Provider The model platform that receives the request after the gateway finishes policy checks, redaction, budget checks, and routing. | perplexity |
Gateway route The configured operation. Only Responses is executable today; embeddings and rerank entries are read-only until their endpoints and admission gates ship. | responses |
Provider lifecycle The approval state for the route/model pairing. Approved aliases are available for governed traffic; preview, deprecated, or disabled statuses signal operational limits. | approved |
Provider route The gateway's internal connector name. Operators use it to trace which provider integration, credentials, defaults, and request mapping are used. | perplexity-sonar |
Underlying model The configured provider model identifier. New-operation profiles are not yet dispatchable. | sonar |
Provider profile Immutable operation profile metadata; changing the model or dimensions requires a new reviewed alias. | Responses |
Default for Use cases where this alias is the automatic choice. A default badge means users do not need to pick this model manually for that workflow. | |
Use cases Governed workflows that may use this alias. This is the business approval boundary, not broad access for every caller. | |
Alias data classes Data categories this alias is approved to process. Labels such as payment.pan or security.jwt describe policy sensitivity, not necessarily what every request contains. | customer.public |
Alias request features Request options callers are allowed to use through this alias. If a feature is not listed here, the gateway should not expose it for this alias. | responses temperature top_p max_output_tokens idempotency_key |
Structured output Whether the route can return data that conforms to a caller-provided schema. This is useful when downstream systems need predictable JSON instead of free-form text. | Not advertised by this provider route |
Provider regions Regions where the provider route is approved to run. Global means the route is not pinned to a single named region in this registry. | global |
Provider retention postures How the provider is allowed to retain request and response data. provider_no_store means the route is intended for no-store/no-training provider handling. | provider_no_store |
Provider request fields Exact fields the provider connector may send downstream. This helps reviewers see which knobs and payload fields can leave the gateway. | idempotency_key input instructions max_output_tokens model store temperature top_p |
Route capabilities Capabilities advertised by the provider route itself. Alias and use-case settings should only expose capabilities the route can actually support. | responses |
Matrix classes The provider feature matrix's raw data-class approval list. It is the route-level policy ceiling that alias-level data classes should fit within. | customer.public |