Troubleshooting

The problems people actually hit, and how to get past them.

A model is missing from the list. Either its key has not been saved yet — models appear once the key for their service is in Settings → Providers — or it has been switched off. The per-model switch under Settings → Models and the provider’s Enabled switch each hide models without deleting anything. See setting up API keys.
A generation fails immediately. Usually the key is wrong. For OpenAI, check that API billing is enabled before trying the request again.
Chat writes back but never makes an image. With no orchestrator key it falls back to keyword detection. Asking more directly helps, and adding an Anthropic, OpenAI or Google key helps more.
Everything feels slow when a batch runs. Lower Concurrent Tasks in Settings.
A layer did not appear in the export. Hidden layers are skipped, and a generate layer that has never been run has no image yet. See exporting.

Still stuck? Email [email protected] and we will help.