Errors

ModelGates uses standard HTTP status codes and returns errors in an OpenAI-compatible format.

Error format

json

{  "error": {    "message": "Invalid API key",    "type": "invalid_request_error",    "code": "invalid_api_key",    "param": null  }}

Code	Status	Meaning
`invalid_api_key`	401	The API key is missing or invalid
`account_frozen`	402	Account is suspended
`balance_too_low`	402	Insufficient credits for this request
`worst_case_too_expensive`	402	Request would exceed the $10 hard ceiling
`concurrency_limit_exceeded`	429	Too many parallel requests (Retry-After: 5)
`rate_limit_per_key`	429	Rate limit hit (Retry-After: 60)
`invalid_request`	400	Malformed request body
`upstream_unavailable`	503	Provider temporarily down
`upstream_timeout`	504	Provider did not respond in time
`internal_error`	500	Unexpected server error

For 429, 503, and 504 errors, we recommend exponential backoff with jitter. The Retry-After header is included when applicable.