Safeguard Docs
API Platform (Console)

Status and support

Check the live status of Safeguard's services, work out what an API Platform error means, and get help from the Safeguard team.

Service status

status.safeguard.sh shows the live state of Safeguard's services:

  • API: the combined platform API, which includes the service that runs the API Platform's gateway;
  • the web apps: Web App, Documentation, Academy, Gold Registry and Trust Center;
  • 90 days of uptime history, and past incidents;
  • notices from the Safeguard team.

Select Subscribe to get status updates by email.

When a request fails

ResponseWhat it meansWhat to do
403 gateway_disabled, or the Console says The inference gateway is not connected for this workspace yetThe API Platform is not switched on for your workspace.Ask Safeguard to switch it on.
401 invalid_api_keyThe key is missing, malformed, unknown, revoked or expired.Check the Authorization: Bearer header, or create a new key.
402 insufficient_creditThe request could cost more than your available credit.Add credit, or lower max_tokens.
402 key_spend_limit_exceeded or 402 account_spend_limit_exceededThe request would pass a monthly spend limit.Raise the limit, or wait for the next month (UTC).
429 rate_limitedToo many requests for this key.Wait for the time in the Retry-After header.
503 model_unavailableThe model can't be served right now, or it doesn't offer zero data retention and your request requires it.Pick another model.
502 upstream_error or 504 upstream_timeoutThe model's provider failed or timed out.Try again.

Every error code is listed on Errors. See Pricing and billing for credit and limits, and Zero data retention for ZDR.

Get help

More on getting help is in Help and support.

What to include

When you ask about a specific request, include:

  • its ID: the response's id or its X-Generation-Id header, which starts with gen-;
  • the time (UTC), the model and the error code.

Never send your API key. Safeguard support doesn't need it.

On this page