model_permaslug | str | ✅ | Permanent slug of the model this endpoint serves. | openai/gpt-4o-2024-08-06 |
provider_slug | str | ✅ | Slug of the upstream provider. | azure |
upstream_model_id | str | ✅ | Model or deployment identifier sent to the upstream provider. | gpt-4o-prod |
http_referer | Optional[str] | ➖ | The app identifier should be your app’s URL and is used as the primary identifier for rankings. This is used to track API usage per application.
| |
x_open_router_title | Optional[str] | ➖ | The app display name allows you to customize how your app appears in OpenRouter’s dashboard.
| |
x_open_router_categories | Optional[str] | ➖ | Comma-separated list of app categories (e.g. “cli-agent,cloud-agent”). Used for marketplace rankings.
| |
idempotency_key | Optional[str] | ➖ | Retry-safe create: a repeated create with the same key from the same organization returns the endpoint the first request created instead of creating another. The endpoint is returned as it is now. Reusing a key with a different request body (model, provider, base URL, upstream model ID, declared ZDR or region, pricing, or declared parameters) is rejected with 422 idempotency_key_reused. Declared parameters are compared as a set, so a retry that reorders or repeats a name replays the endpoint. activate and max_completion_tokens are not compared, and later edits to the endpoint do not affect the comparison. | wayfair-gpt-4o-eastus-2026-09 |
activate | Optional[components.PrivateEndpointActivation] | ➖ | Validate and activate in the same call. On a failed validation the draft is kept and returned with a 422, so fix it and call /validate and /activate instead of creating it again. | { “workspace_id”: “550e8400-e29b-41d4-a716-446655440000” } |
base_url | Optional[str] | ➖ | HTTPS base URL of your deployment. Required unless the provider derives its URL from the BYOK credential (Azure, Amazon Bedrock, Google Vertex). | https://contoso.openai.azure.com |
declared_parameters | List[components.PrivateEndpointDeclaredParameter] | ➖ | Parameters the upstream supports beyond the tested ones. Replaces the earlier declaration; tested parameters stay. Order and repeated names are ignored, but the list holds at most as many entries as there are declarable parameters, repeats included. tool_choice requires tools to be tested or declared. | [ “temperature”, “seed” ] |
declared_region | OptionalNullable[components.CreatePrivateEndpointRequestDeclaredRegion] | ➖ | Attest where this deployment processes data. | us |
declared_zdr | OptionalNullable[bool] | ➖ | Attest that this deployment retains no prompt or completion data. | |
max_completion_tokens | Optional[int] | ➖ | Cap on completion tokens per request. Requests asking for more skip this endpoint, and the default output length stays within it. When omitted, the cap is copied from the public endpoint for the same model and provider, if there is one. | 131072 |
pricing | Optional[components.PrivateEndpointPricing] | ➖ | Negotiated per-token rates reported for requests routed to this endpoint. | { “cache_read”: “0.00000025”, “cache_write”: “0.000003125”, “completion”: “0.00001”, “prompt”: “0.0000025” } |
retries | Optional[utils.RetryConfig] | ➖ | Configuration to override the default retry behavior of the client. | |