RateLimit: Rate limit exceeded - typically HTTP 429. Suggests switching to another provider or waitingNoCredit: Insufficient credits or quota for billing - typically HTTP 402/403. Account has no credits/balance or quota exhausted. Suggests switching to another providerAuthentication: Authentication or authorization failure - typically HTTP 401/403. Usually indicates invalid API key or permissions issueNoCredentials: No valid API credentials/keys configured for any model-vendor combination. This is a configuration error that cannot be resolved by retrying — the user must configure credentialsServiceUnavailable: Service temporarily unavailable - typically HTTP 503. Suggests the service is down or overloadedInternalServerError: Internal server error - typically HTTP 500. Indicates a problem on the provider's sideNetworkError: Network connectivity issues. Connection timeouts, DNS failures, etc.InvalidRequest: Invalid request format or parameters - typically HTTP 400. Usually indicates a structural problem with the request (malformed JSON, invalid syntax)VendorValidationError: Vendor-specific schema or field validation error - typically HTTP 400/422. Field requirements, missing properties, schema validation that may be vendor-specificContextLengthExceeded: Context length exceeded - typically HTTP 400 with context_length_exceeded code. Suggests switching to a model with larger context windowModelError: Model-specific errors. Model not found, model overloaded, etc.Unknown: Unknown or unclassified error
Categorizes error types that can occur during AI model operations. These categories help determine appropriate retry and failover strategies.