[DOCS]

Errors

Every error uses the OpenAI envelope, so SDK error handling works unchanged. The code field is stable and machine-readable. This page lists every code the platform can emit.

ERROR ENVELOPE

{
  "error": {
    "message": "This API key is disabled. Re-enable it from your dashboard.",
    "type": "authentication_error",
    "code": "api_key_disabled"
  }
}

All error codes

HTTP	CODE	TYPE	MESSAGE
401	`missing_api_key`	authentication_error	Missing API key. Pass it as: Authorization: Bearer sk-blaze-...
401	`invalid_api_key`	authentication_error	Invalid API key.
403	`api_key_disabled`	authentication_error	This API key is disabled. Re-enable it from your dashboard.
400	`invalid_request`	invalid_request_error	The request body is invalid.
404	`model_not_found`	invalid_request_error	Unknown model. See GET /v1/models for available models.
404	`model_not_available`	invalid_request_error	This model is not available yet. See blazeinference.com/models.
413	`payload_too_large`	invalid_request_error	Request body exceeds the 5 MB limit.
400	`context_length_exceeded`	invalid_request_error	The request exceeds the model's context window.
429	`quota_exceeded`	rate_limit_error	Daily free quota reached (100 successful requests/day). Resets at midnight UTC.
429	`rate_limited`	rate_limit_error	Too many requests. Slow down and retry shortly.
502	`inference_error`	api_error	Inference failed. This request did not count against your quota. Please retry.
504	`inference_timeout`	api_error	Inference timed out. Please retry.
500	`internal_error`	api_error	Internal error. Please retry.
400	`invalid_otp`	invalid_request_error	Invalid code.
400	`otp_expired`	invalid_request_error	This code has expired or was already used. Request a new one.

Errors during streaming

If a stream fails after it has started, the final SSE event carries the same envelope (typically inference_error or inference_timeout) followed by data: [DONE]. Failed streams never count against your quota.