Message: Generation failed (503 – high demand)¶

What this message actually means¶

When you see a message like:

“Generation failed: This model is currently experiencing high demand.” Error code: 503 – UNAVAILABLE

it can feel like something broke or failed unexpectedly.

But in reality, nothing is wrong with your project or prompt.

What happened is simple: the model is temporarily overloaded and cannot accept your request at that moment.

This is a temporary availability issue, not a failure in your workflow.

Note

Your request was not processed. No generation was completed, and no resources were consumed.

What’s happening behind the scenes¶

When you start a generation, Craftology sends your request to an external model.

Sometimes, due to traffic spikes or limited capacity, the model cannot handle new requests immediately. When that happens, it returns:

503 – UNAVAILABLE

This means:

the service is reachable
but it is currently too busy to respond

Unlike timeouts or workflow errors, this happens before generation even begins.

Why this happens¶

This issue is usually temporary and caused by:

high demand from many users at the same time
limited model capacity
short spikes in usage

It is not related to:

your prompt
your project settings
your workflow

Even simple requests can trigger this message during peak load.

What you should do next¶

Try again after a short wait¶

In most cases, the issue resolves quickly.

Wait a few seconds (or up to a minute) and run the generation again.

Tip

A simple retry is often all it takes.

Retry a few times if needed¶

If the first retry fails, try again after a slightly longer pause.

Spacing out retries gives the system time to recover and avoids adding more load.

Avoid rapid repeated requests¶

Clicking “generate” many times in quick succession can make the issue worse.

Instead:

wait between attempts
retry gradually

This improves your chances of success.

Reduce simultaneous generations¶

If you are running multiple generations at once, they may compete for limited capacity.

For better reliability:

run fewer tasks at the same time
wait for one to start before launching another

Try again later if the issue persists¶

If the error continues for several minutes, it likely means the system is under sustained load.

In that case:

pause for a few minutes
return and try again

Best practice¶

Treat this error as a temporary traffic issue, similar to a busy server.

You don’t need to change anything in your setup — just retry at the right moment.

Tip

Short delays between retries are more effective than repeated rapid attempts.

When to contact support¶

You should reach out if:

the error persists for an extended period
retries never succeed, even after waiting
other types of errors appear alongside it

These cases may indicate a broader service issue.

Review the Reporting an Issue topic for more information.