Error Codes

Name: Chuizi AI
Price: 5 USD

OpenAI-compatible paths return a unified JSON error shape. Anthropic- and Gemini-compatible paths preserve their protocol behavior where practical, while sharing the same authentication, balance, and limit checks.

config.json

json

{
  "error": {
    "message": "Human-readable description",
    "type": "invalid_request_error",
    "code": "validation_error"
  }
}

HTTP	Common code	Meaning	Recommended action
400	`bad_request` / `validation_error`	Invalid body or parameter	Check required fields such as `model` and `messages`
401	`invalid_api_key`	Missing, malformed, or unknown key	Use a valid `ck-` key
402	`insufficient_balance`	Balance cannot cover the estimated request	Top up and retry
403	`model_not_allowed` / `ip_not_allowed` / `daily_limit_exceeded`	Key policy, IP, or daily cap blocked the request	Check API key settings
404	`model_not_found` / `generation_not_found`	Model or generation was not found	Check `/v1/models` or the `gen-` ID
429	`rate_limit_exceeded`	Request rate exceeded limits	Retry with exponential backoff
500	`internal_error`	Gateway internal error	Contact support with `X-Request-Id` if persistent
502	`provider_error`	Upstream returned an error	Retry later
503	`provider_unavailable`	Upstream temporarily unavailable	Retry later or choose another model
504	`provider_timeout`	Upstream timed out	Shorten input/output and retry

Retry-After may appear on some paths or upstream responses, but it is not the current stable public contract. Use exponential backoff in clients.