Could we get a feature to parse and display the "Thinking" process for models that use Chain of Thought.