APIInference

Inference Connections_read

GET
/inference/connections/{connection_id}

Authorization

ApiKeyAuth
headerAuthorizationBearer <token>

Bearer token authentication. Send an API key (a saai_api_-prefixed token) as Authorization: Bearer <token>.

Required permission

view_inference

Path Parameters

connection_id*string
Formatuuid

Response Body

The request has succeeded.

application/json
  1. response

A credential + endpoint an org can use for inference.

id*uuid
Formatuuid
name*string
Length1 <= length <= 255
provider_type*ProviderType

The API a connection speaks. Each type sets its own base URL form and whether an API key is required.

Value in"openai""anthropic"
base_url?string

Absent on a managed connection: the platform's endpoint is operator infrastructure, not the org's to see.

Lengthlength <= 2048
allowed_models*array<string>

Models this connection may be bound to. ["*"] allows all.

has_api_key?boolean

Read-only: whether an API key is stored. The key itself is never returned. Absent on a managed connection, whose credential the org neither sets nor fixes.

managed*boolean

Read-only: true for the platform-provided connection, which the org can't edit or delete.

created_at?string

Absent on a managed connection.

Formatdate-time
updated_at?string

Absent on a managed connection.

Formatdate-time
curl -X GET "https://example.com/inference/connections/497f6eca-6276-4993-bfeb-53cbbbba6f08"
{  "id": "497f6eca-6276-4993-bfeb-53cbbbba6f08",  "name": "string",  "provider_type": "openai",  "base_url": "string",  "allowed_models": [    "string"  ],  "has_api_key": true,  "managed": true,  "created_at": "2019-08-24T14:15:22Z",  "updated_at": "2019-08-24T14:15:22Z"}