The prediction service routes AI inference requests to the appropriate models. Errors from this service typically indicate that a request failed schema validation or violated a business rule, that the requested application type or model version isn’t supported in the current environment, that authentication failed, or that a required model configuration couldn’t be found on the platform.
PS_APP_TYPE_NOT_SUPPORTED
The applicationType value in your request doesn’t match any predictor this deployment supports. Your request couldn’t be completed.
This error occurs when:
The applicationType value is misspelled or uses incorrect capitalization.
The application type is supported in another environment but not in the one you’re calling.
The application type was deprecated or renamed and your request still uses the old name.
How to Troubleshoot
To resolve this error:
Check the applicationType value against the supported application types for the environment you’re calling.
Confirm you’re calling the right environment for the application type you need.
If the application type should be supported in this environment, contact Salesforce Customer Support to confirm its rollout status.
If the issue persists, contact Salesforce Customer Support with the following information.
The trace ID from the error response (instance field, format urn:trace:...)
The error code: PS_APP_TYPE_NOT_SUPPORTED
The applicationType value you submitted
PS_AUTHENTICATION_FAILED
Your request couldn’t be authenticated. The credential was missing, malformed, expired, or not accepted for this endpoint.
This error occurs when:
The authentication header was missing from the request.
The credential was malformed or had expired.
The credential was issued for a different environment or audience than the one you’re calling.
How to Troubleshoot
To resolve this error:
Confirm your request includes the expected authentication header for this endpoint.
If you’re using a token, check whether it has expired and obtain a fresh one if needed.
If you’re using mTLS, confirm that the client certificate is valid and trusted by the service.
Confirm the credential is scoped to the environment you’re calling.
If the issue persists, contact Salesforce Customer Support with the following information.
The trace ID from the error response (instance field, format urn:trace:...)
The error code: PS_AUTHENTICATION_FAILED
The authentication mechanism you’re using (token, mTLS, and so on) and the environment you called
PS_AUTOML_VERSION_NOT_SUPPORTED
The AutoML version in your request isn’t supported by this deployment. Your request couldn’t be completed.
This error occurs when:
The AutoML version string is misspelled or uses the wrong format.
The AutoML version isn’t yet rolled out in this environment, or has been retired.
Your client is configured to use an AutoML version that’s no longer available.
How to Troubleshoot
To resolve this error:
Check the AutoML version against the supported versions for the environment you’re calling.
Update your client to a currently supported AutoML version.
If the version should be supported in this environment, contact Salesforce Customer Support to confirm its rollout or retirement status.
If the issue persists, contact Salesforce Customer Support with the following information.
The trace ID from the error response (instance field, format urn:trace:...)
The error code: PS_AUTOML_VERSION_NOT_SUPPORTED
The AutoML version value you submitted
PS_BAD_REQUEST
Your request was rejected before it reached a model because it failed validation or violated a business rule.
This error occurs when:
A required field was missing or empty.
A field value was the wrong type or outside the allowed range for the application type you’re calling.
The request combined fields in a way that the targeted predictor doesn’t accept.
The request body wasn’t valid JSON or didn’t match the API schema.
How to Troubleshoot
To resolve this error:
Check the detail field in the error response. It identifies which field or rule caused the rejection.
Validate your request against the API schema for the application type you’re calling.
Compare your request against a known-good example for the same application type and correct any differences.
If the issue persists, contact Salesforce Customer Support with the following information.
The trace ID from the error response (instance field, format urn:trace:...)
The error code: PS_BAD_REQUEST
A redacted copy of the request body
PS_ML_VERSION_NOT_SUPPORTED
The ML model or runtime version in your request isn’t supported by this deployment. Your request couldn’t be completed.
This error occurs when:
The version string is misspelled or uses the wrong format.
The version isn’t yet rolled out in this environment, or has been retired.
Your client is pinned to a model version that’s no longer available.
How to Troubleshoot
To resolve this error:
Check the version against the supported ML model and runtime versions for the environment you’re calling.
Update your client to a currently supported version.
If the version should be supported in this environment, contact Salesforce Customer Support to confirm its rollout or retirement status.
If the issue persists, contact Salesforce Customer Support with the following information.
The trace ID from the error response (instance field, format urn:trace:...)
The error code: PS_ML_VERSION_NOT_SUPPORTED
The ML version value you submitted
PS_REQUEST_TOO_LARGE
Your request exceeded the maximum size the prediction service accepts. Your request couldn’t be completed.
This error occurs when:
The input batch contained more records than the service accepts in a single call.
A single record contained an unusually large text or data payload.
A metadata field in the request carried more content than expected.
How to Troubleshoot
To resolve this error:
Reduce the size of the request payload.
If you’re sending a batch, split it into smaller batches and send them as separate requests.
If a single record is too large, trim its content before submitting.
If the issue persists, contact Salesforce Customer Support with the following information.
The trace ID from the error response (instance field, format urn:trace:...)
The error code: PS_REQUEST_TOO_LARGE
The approximate size of the payload you were sending
PS_RESOURCE_NOT_FOUND
This error code indicates an unexpected system error and not an issue with your configuration. If the issue persists, contact Salesforce Customer Support with the following information.
The trace ID from the error response (instance field, format urn:trace:...)
The error code: PS_RESOURCE_NOT_FOUND
Any context that might help with the investigation, such as when the error occurred, what you or your end user were doing at the time, and whether you’re able to reproduce it
j