Platform API Errors and Limits

Platform API errors use a different body shape from model endpoint errors: the HTTP status carries the result, and the JSON body always includes success and, on failure, a machine-readable code.

Response format

On success, HTTP 200 is returned:

{
  "success": true,
  "message": "",
  "data": {}
}

On failure, the real HTTP status code is used and the body includes a machine-readable code:

{
  "success": false,
  "message": "Invalid Management Key",
  "code": "invalid_token"
}

message is localized and may change between versions. Use the HTTP status code and code in program logic.

Error codes

HTTP	code	Scenario
400	invalid_request	Invalid parameters, such as amount out of range, missing or overlong name on create, status outside 1 / 2, enabling a Key that is still expired or exhausted, invalid expiration time, invalid pagination or time range, or an unencoded ; in the query
401	invalid_token	Missing Authorization header, malformed Key, or Key not found or deleted
401	token_expired	The Management Key has expired
401	token_revoked	The Management Key was revoked (legacy Keys only; newly created Keys can only be deleted)
403	user_disabled	The Management Key owner’s account is disabled
403	insufficient_scope	The Management Key does not have the permission this endpoint requires
404	not_found	The API Key does not exist or does not belong to you (not distinguished); or the path does not exist under /api/open/v1
405	method_not_allowed	The path exists but does not support that HTTP method
429	rate_limited	Per-Management-Key rate limit exceeded
500	internal_error	Server error; internal details are not returned

Unknown paths under /api/open/v1 return 404 and a wrong HTTP method returns 405, both with the unified error body above; use code to tell “retry with another method” from “wrong path or parameters”.

Rate limits

Range	Limit	Applies to	Response when exceeded
Per Management Key	120 requests/minute	All Platform API endpoints	429 with code rate_limited
Per user	30 requests/minute (default)	Account logs, Export logs, Usage, Usage by model, Usage by Key	429; the response body may be empty
Per client IP	180 requests/180 seconds (default)	Shared across all /api endpoints	429
Per client IP	100 requests/20 minutes (default)	Sensitive endpoints such as Reveal an API Key	429; the response body may be empty

Handle rate limiting by HTTP status code. When you receive 429, slow down and retry later.

Model endpoints use a different error body and separate limits; see Common Errors.