I'm using the Chatbox application with the Gemini 2.0 model. When the AI attempts to generate tables or structured content (such as comparing technical-economic analysis methods versus investment control methods), the application freezes and produces corrupted output - displaying numerous repetitions of the Chinese character "在" (zài) across the screen, making the response completely unusable. I've provided a screenshot as evidence showing the abnormal state when the AI tries to render tables. This issue occurs consistently whenever tables are requested, severely limiting the application's functionality. The problem does not appear with simple text responses, only when structured content like tables is generated.