List Providers

List providers currently configured and available to your key.

GET/v1/providers

Query Parameters

Name	Type	Required	Description
include_health	string		Set to "true" to include health and latency snapshot for each provider. When omitted, status is "unknown" and the health field is not returned.
type	string		Filter providers by type. Valid values: "anthropic", "openai", "bedrock", "azure", "vertex-ai". Invalid types return an empty data array.

Request Example

bash

curl https://api.therouter.ai/v1/providers \
  -H "Authorization: Bearer $THEROUTER_API_KEY"

Response

json

{
  "data": [
    {
      "id": "openai-api",
      "name": "OpenAI",
      "type": "openai",
      "status": "unknown",
      "models_count": 12
    },
    {
      "id": "bedrock-us-east-2",
      "name": "AWS Bedrock (us-east-2)",
      "type": "bedrock",
      "status": "unknown",
      "models_count": 8
    }
  ]
}

Response with Health

When include_health=true is passed, each provider includes a health snapshot and a derived status.

bash

curl "https://api.therouter.ai/v1/providers?include_health=true" \
  -H "Authorization: Bearer $THEROUTER_API_KEY"

json

{
  "data": [
    {
      "id": "openai-api",
      "name": "OpenAI",
      "type": "openai",
      "status": "healthy",
      "models_count": 12,
      "health": {
        "latency_p50": 45,
        "latency_p99": 45,
        "error_rate": 0.0,
        "last_checked_at": "2026-02-26T10:00:00.000Z"
      }
    },
    {
      "id": "bedrock-eu-west-1",
      "name": "AWS Bedrock (eu-west-1)",
      "type": "bedrock",
      "status": "unhealthy",
      "models_count": 0,
      "health": {
        "latency_p50": 0,
        "latency_p99": 0,
        "error_rate": 1.0,
        "last_checked_at": "2026-02-26T10:00:00.000Z"
      }
    }
  ]
}

Status Derivation

When include_health=true, each provider's status is derived from its health check results using these thresholds:

Name	Type	Required	Description
healthy	status	Required	Error rate < 5% AND latency < 500ms.
degraded	status	Required	Error rate < 20% AND latency < 2000ms (but not healthy).
unhealthy	status	Required	Error rate >= 20% OR latency >= 2000ms, or health check failed.
unknown	status	Required	Returned when include_health is not set.

Response Fields

Name	Type	Required	Description
id	string	Required	Provider identifier (e.g., "openai-api", "bedrock-us-east-2").
name	string	Required	Human-readable display name (e.g., "AWS Bedrock (us-east-2)").
type	string	Required	Provider type: "anthropic", "openai", "bedrock", "azure", or "vertex-ai".
status	string	Required	One of "healthy", "degraded", "unhealthy", or "unknown". Returns "unknown" when include_health is not set.
models_count	integer	Required	Number of models routed through this provider as primary.
health	object		Health snapshot. Only present when include_health=true. Contains latency_p50, latency_p99 (ms), error_rate (0.0-1.0), and last_checked_at (ISO 8601).

Notes

Without include_health=true, the endpoint returns instantly from in-memory data. With health checks enabled, response time depends on provider latency (up to 2s worst case). Provider health can be used to explain routing fallback behavior.