Has anybody been experiencing the "Fail to parse structured output" error from LLM nodes?

All workflows that include LLM calling for a structured output started to throw this error msg this evening. All workflows remain unchanged and they were working perfectly before, didn’t see any suspicioud reports from the model provider(Google AI studio) side.

The plugin execution on Dify Cloud has seemed unstable for about eight hours.
Sometimes the app responds, and sometimes it doesn’t.


Yes! I experienced that too! if fails around 50% in my conversation. try to debug for many time but just dont know what happened! So annoying

I’m failing about 70% of the time. Anybody have an idea on when they are going to fix this?

It looks like the issue was resolved around 6:00 AM (UTC) on March 23.

1 Like