This issue started suddenly today. Not only in responses to questions, but also when generating documents or modifying code, any content that includes Korean text ends up with broken encoding.
Has anyone experienced this issue, or is there a way to fix it?
Steps to Reproduce
Enable Auto Mode
Ask a question or generate code including Korean text
The output shows broken encoding
Hey, thanks for the report. I can see the screenshot shows broken text encoding.
This is a known issue with Auto Mode. The models behind it don’t handle non Latin text correctly (Korean, Vietnamese, Chinese, etc.). We’ve received a few similar reports today.
As a workaround, try picking a specific model instead of Auto (for example, Claude 4.6 Sonnet or GPT-5) in the model selector. This should fix the encoding issue.
The team is aware. Your report helps us prioritize this. Let me know if the workaround doesn’t help.
Hey @boyism80,
Just checking in — are you still experiencing the encoding issue with Korean text in Auto mode? Recent updates should have improved this. If you’re still seeing it after updating to the latest version, let me know and I’ll dig deeper.
I’m currently using it without any issues. Occasionally, a small portion of text appears garbled, but it’s not significant enough to affect understanding the context. Overall, it’s working very well. Thank you.
