API REFERENCE

Errors & limits

{
  "error": {
    "code": "invalid_request",
    "message": "Check the request fields.",
    "details": [
      {
        "path": "message",
        "message": "Message cannot be empty."
      }
    ]
  }
}

STATUS	MEANING
400 / 415 / 422	Invalid JSON, unsupported content type, or invalid fields. Unknown fields are rejected.
401	Missing or incorrect API key.
402	The free or monthly request allowance is exhausted. Requests are blocked until an upgrade or the paid plan’s next billing period; there are no automatic overage charges.
409	The same Idempotency-Key is already running or was used with different input.
408	The caller cancelled the request.
413	Request body exceeds 64 KiB.
429	Service or inference provider rate limit. Read `Retry-After`.
502 / 503 / 504	Provider error, missing server configuration, or the 1.5-second upstream deadline expired.

Each account can make 300 requests per minute, shared across its keys. Message, personality, and conversation text, previous reaction IDs, and reaction options together are limited to 12,000 UTF-8 bytes. Encoded upstream requests are also limited to 20,000 bytes; shorten descriptions or other input if you receive input_too_large.

Latency depends on payload size, network conditions, and provider load. The 1.5-second inference deadline does not include all request processing and network time.