We're seeing elevated error rates on AI model requests, which may cause message failures or timeouts when generating or editing projects.

We're actively investigating and will post an update as soon as we know more.

Retrying a failed message may succeed in the meantime.