Mark a provider's discovered models as reviewed

POST

/v1/providers/{id}/models/review

const url = 'https://app.everruns.com/api/v1/providers/example/models/review';
const options = {method: 'POST'};

try {
  const response = await fetch(url, options);
  const data = await response.json();
  console.log(data);
} catch (error) {
  console.error(error);
}

curl --request POST \
  --url https://app.everruns.com/api/v1/providers/example/models/review

Clears the is_new flag on the provider’s discovered models: only models discovered after this call, and still disabled, read back as new.

Parameters

Path Parameters

required

string

Provider ID (prefixed, e.g., prov_…)

Responses

200

Provider with its review time updated

application/json

LLM Provider entity (API keys never exposed) This is the persisted provider entity, separate from the runtime provider trait.

object

api_key_set

required

Whether an API key is configured. The key itself is never returned.

boolean

base_url

Custom base URL for self-hosted / proxied providers. None means use the provider’s default endpoint.

string | null

created_at

required

Timestamp when this provider was created (RFC 3339).

string format: date-time

required

Prefixed public identifier. See ID Schema.

string

last_synced_at

Timestamp of the most recent successful model sync from the provider’s API (RFC 3339).

string | null format: date-time

managed

required

Whether this provider is host-managed (EVE-810). A managed provider is provisioned by the host/embedder; the OSS API rejects tenant PATCH/DELETE on it (403). Read-only to org admins. Defaults to false.

boolean

models_reviewed_at

When someone last reviewed this provider’s discovered models (RFC 3339). A discovered model created later that is still disabled is reported as is_new on the model. None means never reviewed.

string | null format: date-time

name

required

Human-readable provider name. Safe to render in user-facing messages.

string

provider_type

required

Provider implementation type (OpenAI, Anthropic, Gemini, etc.).

string

request_options

One of:

object
null

Extra headers and diagnostics options applied to every request sent to this provider. None when the org configured nothing.

object

cache_diagnostics

Ask the provider to explain unexpected prompt-cache misses. Honored by drivers with a diagnostics protocol (today: Anthropic’s cache-diagnosis beta); ignored elsewhere.

boolean

headers

Extra HTTP headers added to every request to this provider. They override the driver’s and the connection’s own headers by name, but connection-level headers (host, content-length, …) are ignored.

Array<object>

One extra HTTP header sent with every request to a provider connection.

object

name

required

Header name, e.g. x-gateway-tenant.

string

value

required

Header value, sent verbatim.

string

status

required

Current lifecycle status of this provider.

string

Allowed values: active disabled

trace

One of:

object
null

Resolved trace/observability link configuration: the driver’s default templates overlaid with this provider’s stored overrides. None when the driver exposes no dashboard and the org configured nothing.

object

enabled

required

Whether trace links should be shown for this provider. Defaults to false: vendors typically do not retain trace content unless logging is explicitly enabled, so the org opts in once that is set up.

boolean

generation_url_template

URL template for a single generation’s trace, e.g. "https://openrouter.ai/logs?id={response_id}".

string | null

session_url_template

URL template for a session’s grouped trace, e.g. "https://openrouter.ai/logs".

string | null

updated_at

required

Timestamp when this provider was last updated (RFC 3339).

string format: date-time

Example

{
  "id": "provider_01933b5a00007000800000000000001",
  "status": "active"
}

400

Invalid provider ID

403

Caller cannot manage providers

404

Provider not found