Hi @NWD Thank you for the post. I checked the request and the screenshot.
From what I can see, this does not look like another user’s data was included in your request. The request was handled under your own account context, and I do not see evidence of a different user’s chat, files, or workspace being attached to it.
What appears to have happened is that the model produced corrupted or unrelated text in the Thinking view. That can look alarming because it may switch languages or mention topics unrelated to your project, but in this case, the screenshot looks like degraded model output rather than data from another user’s session.
If you see it again, feel free to grab the new request ID and a screenshot if possible. Starting a fresh chat is the best immediate workaround if the conversation starts drifting or producing unrelated text.
I’m sorry but it’s really not making any sense that totally non related outputs generated, really feel wrong especially since it’s counting as tokens generated from our sessions, so you state that the model is generating garbage for 1/3 of it’s output, so users are charged for 1/3 of output tokens that are unrelated to the work that should be performed ?
