Skip to content
darkseek
API reference

Responses

Use the backend’s Responses-format generation endpoint.

POST/v1/responses

Request

Use model, input and max_output_tokens. Background execution is disabled. Storage/retrieval, hosted tools and long-lived response IDs are not promised by the gateway.

{
  "model":"darkseek-v4.1",
  "input":"Reply with a short greeting.",
  "max_output_tokens":128,
  "stream":false
}

Verification boundary

A tiny Responses request passed on the existing authenticated backend. Account quotas, gateway streaming usage reconciliation and the public Responses path still need end-to-end launch verification. An internal backend probe is not proof of public availability.