Resources
Inference errors
Resources
Inference errors
This page lists common error codes encountered during inference requests using the Fireworks API, their meanings, and potential resolutions.
Error codes
Below is a table of common status codes and their associated messages for inference-related API requests.
| Error Code | Error Name | Possible Issue(s) | How to Resolve |
|---|---|---|---|
400 | Bad Request | Invalid input or malformed request. | Review the request parameters and ensure they match the expected format. |
401 | Unauthorized | Invalid API key or insufficient permissions. | Verify your API key and ensure it has the correct permissions. |
402 | Payment Required | User’s account is not on a paid plan or has exceeded usage limits. | Check your billing status and ensure your payment method is up to date. Upgrade your plan if necessary. |
403 | Forbidden | The model name may be incorrect, or the model does not exist. This error is also returned to avoid leaking information about model availability. | Verify the model name on the Fireworks site and ensure it exists. Double-check the spelling of the model name in your request. |
404 | Not Found | The API endpoint is incorrect, or the resource path is invalid (e.g., a user tried accessing /v1/foobar instead of a valid endpoint). | Verify the URL path in your request and ensure you are using the correct API endpoint as per the documentation. |
405 | Method Not Allowed | Using an unsupported HTTP method (e.g., using GET instead of POST). | Check the API documentation for the correct HTTP method to use for the request. |
408 | Request Timeout | The request took too long to complete, possibly due to server overload or network issues. | Retry the request after a brief wait. Consider increasing the timeout value if applicable. |
412 | Precondition Failed | This error occurs when attempting to invoke a LoRA model that failed to load. The final validation of the model happens during inference, not at upload time. | Check the body of the request for a detailed error message. Ensure the LoRA model was uploaded correctly and is compatible. Contact support if the issue persists. |
413 | Payload Too Large | Input data exceeds the allowed size limit. | Reduce the size of the input payload (e.g., by trimming large text or image data). |
429 | Over Quota | The user has reached the API rate limit. | Wait for the quota to reset or upgrade your plan for a higher rate limit. |
500 | Internal Server Error | This indicates a server-side code bug and is unlikely to resolve on its own. | Contact Fireworks support immediately, as this error typically requires intervention from the engineering team. |
502 | Bad Gateway | The server received an invalid response from an upstream server. | Wait and retry the request. If the error persists, it may indicate a server outage. |
503 | Service Unavailable | The service is down for maintenance or experiencing issues. | Retry the request after some time. Check for any maintenance announcements. |
504 | Gateway Timeout | The server did not receive a response in time from an upstream server. | Wait briefly and retry the request. Consider using a shorter input prompt if applicable. |
520 | Unknown Error | An unexpected error occurred with no clear explanation. | Retry the request. If the issue persists, contact support for further assistance. |
Troubleshooting tips
If you encounter an error not listed here, try the following:
- Review the API documentation for the correct usage of endpoints and parameters.
- Check the Fireworks status page for any ongoing service disruptions.
- Contact support at support@fireworks.ai for further assistance.
This will provide additional insights into any issues encountered.
Need more help?
If you continue to experience issues, please reach out on our Discord channel.