OA
OneAI API
Docs
requestIderror coderetryable

Reference: Errors

How OneAI reports validation, policy, and provider failures.

Failure shape

Use requestId and code to support customers and debug provider behavior.

{
  "success": false,
  "requestId": "req_...",
  "error": "Max cost exceeded",
  "code": "LLM_COST_LIMIT_EXCEEDED",
  "retryable": false
}

Common codes

These codes are designed to be customer-safe and operationally useful.

INVALID_API_KEY

The key is missing, revoked, or not found.

MODELS_FORBIDDEN

The current key or plan cannot access model registry data.

MODEL_HEALTH_FORBIDDEN

Health checks require an admin key.

LLM_COST_LIMIT_EXCEEDED

The request exceeded the configured maxCostUsd policy.

UPSTREAM_CONFIG_ERROR

Provider key, base URL, or model config is missing.

UPSTREAM_RATE_LIMITED

Provider returned a retryable rate limit.