Description:
Currently, the model's thought process is collapsed by default and not displayed. However, the majority of the reasoning steps are meaningful, and I find myself manually expanding them every time the model starts inferencing. I've observed that most web-based AI tools in the market default to showing the thought process expanded.

Proposed Change:
Could we consider changing the default behavior to expand the thought process? This would align with common practices in similar tools and reduce the need for repetitive manual actions.

Example:
https://chat.qwen.ai/
https://chat.deepseek.com/