Skip to main content

LLM provider alerts

Glean sends an LLM provider alert when the endpoint for your LLM provider (Azure OpenAI, OpenAI, Google Vertex AI, or Amazon Bedrock) returns repeated errors. The email identifies the affected endpoint and the distribution of HTTP status codes it returned.

This alert appears as LLM provider errors on the Alerts page and is active whenever you use your own LLM provider. For recipients, sender details, and other alert settings, see About admin alerts.

Why you received this alert

Glean calls your LLM provider's endpoint to power Glean Assistant. Because Glean doesn't have administrative access to your provider, you need to resolve these errors on the provider side. Until the errors stop, some or all of Glean Assistant's functionality is affected.

Resolve the errors

The HTTP status codes in the alert email indicate what's going wrong:

Status codeMeaning
401Invalid or expired API key
403Access denied, such as an IP allowlist, firewall, or permissions issue
404Deployment or model not found
429Rate limit exceeded because your provisioned capacity is too low
500Server or gateway error, often a provider outage
TimeoutEndpoint is slow or unreachable, or a gateway is adding latency

For step-by-step resolution for each status code, see Troubleshoot LLM provider errors.

Once the errors are resolved, you stop receiving alert emails. No further action is required.