Fix a failed or stuck generation
Last updated: October 11, 2026
A generation ended with a message on its tile instead of a result, or it has been running far longer than usual. Find the message you see below; each one points to the input, the model, your account or Layer itself.
The color of the message tells you whose move it is. Orange means the request was understood and declined, or an input cannot be used: change something and run it again. Red means Layer or a service it depends on failed: you did nothing wrong, and trying again shortly is the fix. A run that fails without producing anything is not charged; the Creative Units held for it while it ran are released.
Most likely causes
The run failed with "Something went wrong" or "Generation failed"
A red message with no specific reason, such as Something went wrong. Please try again. or Generation failed, is a fault on Layer's side or the model provider's. Run the same request again. If it fails a second time, try another model from the model menu. If every model fails, check for an incident (see the last section) before changing your prompt.
The run never finishes
A running tile shows how long the generation has been going, and some tools take minutes rather than seconds. Every run has a time limit: one that exceeds it is marked failed on its own, so you are not left waiting forever. To stop it yourself, hover the tile and click Cancel generation, then run it again or switch model.
The result was blocked by content moderation
These appear in orange:
Rejected by our content safety system
The model provider refused the prompt or an input image under its own policy. Rephrase the prompt, remove the input that triggered it, or try another model, since providers draw the line in different places.
Your prompt may contain intellectual property references that cannot be processed
Remove named characters, brands or franchises from the prompt and describe the look instead.
This request was blocked before it reached the model
Your workspace or project has a rule that stopped the request, and the message names it. Change the request to fit the rule, or ask a workspace admin about it.
The model returned no result
The provider produced nothing and gave no reason, which usually means it quietly rejected the prompt or inputs. Rephrase the prompt before running it again; the same request tends to fail the same way.
An input was rejected
File too large
Appears when you upload: the message gives the maximum size for that kind of file. Compress or resize the file and upload it again. See Supported file types and size limits for every limit and accepted format.
only .mp4 and .mov videos are supported
Appears when you drop a video into a tool or the chat: convert the clip to MP4 or MOV first.
Image is too small, too large or too long
Image is too small., Image is too large., Video is too long., Audio is too long. and similar: the model has its own limits on input size and length. Resize or trim the input, or choose a model that accepts it.
This model accepts a prompt of at most
The message ends with a number of characters: each model sets its own prompt length. Shorten the prompt to the number the message gives, or switch to a model that takes longer prompts.
One or more inputs are invalid for this model
One or more inputs are invalid for this model. Check your files and settings, then try again. or This model can't run with one of the chosen inputs or settings.: a setting or attached file is not one this model supports. Remove the last input you added, or try another model.
You are out of Creative Units
When Layer sees the shortfall before the run starts, the top-up dialog opens. If you can buy units, pick a pack; once the balance updates, click Resume generation to run the request you were stopped on. If you cannot, the dialog says Ask a workspace admin to add more. When the shortfall shows up during the run, the tile reads Not enough creative units to run this generation. Top up your workspace and try again. A warning in the chat and tool panels (You're out of creative units.) tells you before you run anything.
We couldn't verify your workspace balance. Try again shortly. is different: Layer could not read your balance at that moment. Wait a minute and run it again.
The model is unavailable or busy
A service we rely on is unavailable
The model's provider is down or overloaded. Try again in a few minutes, or use another model in the meantime.
Layer is briefly unavailable
Layer itself is restarting or under load. Wait and try again.
Please retry after a while
A message ending Please retry after a while. means you started more generations in a short window than your account allows. Wait a few minutes before starting more.
That model is no longer available
has been disabled in this workspace or That model is no longer available. A workspace admin turned the model off, or it was retired. Choose a different model, or ask an admin to enable it.
Narrow it down
Check for an incident. When Layer has an ongoing incident or maintenance, a banner shows in the sidebar. You can also open Help & Resources > Status to see the status page.
Run it again unchanged. A red failure that succeeds the second time was transient.
Try another model with the same prompt and inputs. If it works, the first model or its provider is the problem.
Remove inputs one at a time. If the model works with a plain prompt, one of your files or settings is the problem.
Check your account. Out of Creative Units, a model disabled for your workspace, or a workspace rule all point to your account or an admin, not to the request.
Still stuck?
Contact support from Help & Resources > Talk To Us in Layer. Include the session link, the message exactly as it appears on the tile, the model you used, and roughly when it happened. Contact support with the right details lists everything that helps.