Description
Two related issues occur during AI response generation:
- The Stop button does not work — clicking the stop/cancel button (■) has no effect; the AI continues generating regardless
- The chat history and AI output are not visible during generation — the chat area remains completely black/empty while the response is being streamed
Observed Behavior
- A message is sent and the AI starts processing (timer shows e.g. "1m 13s", progress shows 3%)
- The chat area is completely blank — neither the user's message nor any AI output is visible
- The stop button (■) in the bottom right is shown but clicking it does not cancel the generation
- The input field shows "Queue a message..." suggesting the system is locked in a generating state
- The only visible element is the user's last message bubble in the top right corner
Expected Behavior
- The chat history should remain fully visible during generation
- Streamed AI output should appear in real-time as it is generated
- The stop button should immediately cancel the ongoing generation
- After stopping, the partially generated response should remain visible (or be clearly marked as interrupted)
Why This Matters
- Users have no way to interrupt long or incorrect responses
- The blank screen during generation is disorienting — there is no feedback that anything is happening except the timer and progress percentage
- Makes the app feel unresponsive and broken during longer tasks
Environment
- OpenGUI Desktop Interface
- OS: Linux
- Model: GPT-5.4, Build mode
Description
Two related issues occur during AI response generation:
Observed Behavior
Expected Behavior
Why This Matters
Environment